source: Validate External Links/validate_external_links.sh@ 1118

#!/bin/bash

# Validate External Links by Iritscen
# Provided with a list of external links found in the OniGalore wiki, this script validates them.
# The resulting logs are produced in three formats: TXT (for easy diffing with an earlier log), RTF
# (for reading as a local file with clickable links), and HTML (for uploading as a web page).
# Call script with "--help" argument for documentation. Also see Read Me First.rtf for critical notes.
# Recommended rule:
# |----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----|

# Set separator token to newline
IFS="
"

### GLOBALS ###
# Settings -- these will be changed from their defaults by the arguments passed in to the script
LINKS_URL=""        # use 'curl' to download file with links from this location (can be file://)
EXCEPT_URL=""       # ditto above for file with exceptions to NG results
OUTPUT_DIR=""       # place reports and all other output in a folder inside this existing folder
RECORD_OK_LINKS=0   # record response code to the log even when it's a value in OK_CODES
SUGGEST_SNAPSHOTS=0 # query the Internet Archive for a possible snapshot URL for each NG page
TAKE_PAGE_SHOT=0    # take a screenshot of each OK page
CHROME_PATH=""      # path to a copy of Google Chrome that has the command-line screenshot feature
URL_START=1         # start at this URL in LINKS_FILE (1 by default)
URL_LIMIT=0         # if non-zero, stop at this URL in LINKS_FILE
UPLOAD_INFO=""      # path to a file on your hard drive with the login info needed to upload a report

# Fixed strings -- see the occurrences of these variables to learn their purpose
AGENT="Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.122 Safari/537.36 OPR/67.0.3575.53"
ARCHIVE_API="http://archive.org/wayback/available"
ARCHIVE_GENERIC="https://web.archive.org/web/*"
ARCHIVE_OK_CODES="statuscodes=200&statuscodes=203&statuscodes=206"
CHROME_SCREENSHOT="screenshot.png"
CURL_CODES="http://iritscen.oni2.net/val/curl_codes.txt"
EXPECT_SCRIPT_NAME="val_expect_sftp.txt"
HTTP_CODES="http://iritscen.oni2.net/val/http_codes.txt"
MY_WIKI_PAGE="http://wiki.oni2.net/User:Iritscen"
THIS_DIR=$(cd $(dirname $0); pwd)
WORKING_DIR=$(pwd)
WIKI_PATH="wiki.oni2.net"

# These are parallel arrays of the IDs and names of OniGalore's current namespaces
declare -a NS_IDS=(-2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 100 101 102 103 104 105 108 109 110 111)
declare -a NS_NAMES=("Media" "Special" "Main" "Talk" "User" "User_talk" "OniGalore" "OniGalore_talk" "File" "File_talk" "MediaWiki" "MediaWiki_talk" "Template" "Template_talk" "Help" "Help_talk" "Category" "Category_talk" "BSL" "BSL_talk" "OBD" "OBD_talk" "AE" "AE_talk" "Oni2" "Oni2_talk" "XML" "XML_talk")

# These arrays tell the script which suffixes at the ends of URLs represent files and which are pages.
# This determines whether the script tries to take a screenshot of the URL or just gets its HTTP code.
declare -a HTTP_FILES=(txt zip wmv jpg png m4a bsl rar oni mp3 mov ONWC vbs TRMA mp4 doc avi log gif pdf dmg exe cpp tga 7z wav east BINA xml dll dae xaf fbx 3ds blend flv csv)
declare -a HTTP_TLDS_AND_PAGES=(com net org uk ru de it htm html php pl asp aspx shtml pgi cgi php3 x jsp phtml cfm css action stm js)

# These arrays tells us which HTTP response codes are OK (good), which are RD (redirections), and which
# are NG (no good). Pages that return OK codes will be screenshotted. Remember to update http_codes.txt
# if you add a new code.
declare -a OK_CODES=(200 401 405 406 501)
declare -a RD_CODES=(301 302 303 307 308)
declare -a NG_CODES=(000 403 404 410 500 503)

# Characters not allowed in a URL. Curly braces are sometimes used on the wiki to build a link using
# transcluded text, and if the transclusion fails, then the braces show up in the URL
ILLEGAL_CHARS="{ }"

# The shortest URL possible, used for sanity-checking some URLs: http://a.co
MIN_URL_LENGTH=11

# These are parallel arrays giving the prefixes that can be used in place of normal external links to
# some wikis and other sites
declare -a INTERWIKI_PREFIXES=(commons metawikimedia mw wikibooks wikidata wikimedia wikinews wikiquote wikisource wikispecies wikiversity wikivoyage wikt wp)
declare -a INTERWIKI_DOMAINS=(commons.wikimedia.org meta.wikimedia.org mediawiki.org wikibooks.org wikidata.org wikimediafoundation.org wikinews.org wikiquote.org wikisource.org species.wikimedia.org wikiversity.org wikivoyage.org wiktionary.org wikipedia.org)

# Variables for keeping track of main loop progress and findings
LINK_NUM=0
EI_LINKS=0
IW_LINKS=0
OK_LINKS=0
RD_LINKS=0
NG_LINKS=0
SKIP_UNK_NS=0
SKIP_JS_PAGE=0
SKIP_BAD_URL=0
SKIP_NON_ASCII=0
SKIP_UNK_SUFFIX=0
SKIP_UNK_CODE=0
SKIP_EXPECT_NG=0
SKIP_EXPECT_EI=0
SKIP_EXPECT_IW=0
FILE_LINKS=0
PAGE_LINKS=0
SKIPPED_HEADER_ROW=0
FINISHED_LIST="no"
START_RUN=0
END_RUN=0


### HELP ###
# A pseudo-man page. Here is the 80-character rule for the page text:
# 234567890123456789012345678901234567890123456789012345678901234567890123456789
function printHelp()
{
  cat << EOF

NAME
       Validate External Links

SYNOPSIS
       validate_external_links.sh --help
       validate_external_links.sh --links URL --output DIR [--exceptions URL]
          [--record-ok-links] [--suggest-snapshots] [--take-screenshots FILE]
          [--start-url NUM] [--end-url NUM] [--upload FILE]

DESCRIPTION
       This script parses a list of external links found in the OniGalore wiki
       (which is dumped by the Oni2.net domain periodically in a particular
       format), validates them using the Unix tool 'curl', and produces a report
       of which links were "OK" (responded positively to an HTTP query), which
       were "RD" (responded with a 3xx redirect code), which could be "IW"
       (interwiki) links, which are "EI" (external internal) links and could be
       intrawiki links, and which were "NG" (no good; a negative response to the
       query). This report can then be automatically uploaded to the location of
       your choice. The script can also suggest Internet Archive snapshots for
       "NG" links, and take screenshots of "OK" links for visual verification by
       the reader that the page in question is the one intended to be displayed.

       You must pass this script the URL at which the list of links is found
       (--links) and the path where the directory of logs should be outputted
       (--output). All other arguments are optional.

OPTIONS
       --help                  Show this page.
       --links URL             (required) URL from which to download the CSV
                               file with external links. Note that this URL can
                               be a local file if you supply a file:// path.
       --output DIR            (required) Unix path to directory in which Val
                               should place its reports.
       --exceptions URL        In order to remove links from the report which
                               Val finds an issue with, but which you regard as
                               OK, list those desired exceptions in this file.
                               See the sample file exceptions.txt for details.
                               Note that this URL can point to a local file if
                               you supply a file:// path.
       --record-ok-links       Log a link in the report even if its response
                               code is "OK".
       --suggest-snapshots     Query the Internet Archive for a possible
                               snapshot URL for each "NG" page.
       --take-screenshots FILE Call the Google Chrome binary at this path to
                               take screenshots of each "OK" page.
       --start-url NUM         Start at this link in the links CSV file.
       --end-url NUM           Stop at this link in the links CSV file.
       --upload FILE           Upload report using the credentials and path
                               given in this local text file. See sftp_login.txt
                               for template.

BUGS
       The script cannot properly parse any line in the external links file
       which contains a comma in the name of the wiki page containing a link.
       Commas in the link itself are not an issue.
EOF
}


### SETUP ###
# If first argument is a help request, or if nothing was passed in at all, print help page and quit
if [ "$#" -eq 0 ] || [ "$1" == "--help" ]; then
  printHelp | less
  exit 0
fi

# Parse arguments as long as there are more arguments to process
while (( "$#" )); do
   case "$1" in
      --links )             LINKS_URL="$2";                     shift 2;;
      --exceptions )        EXCEPT_URL="$2";                    shift 2;;
      --output )            OUTPUT_DIR="$2";                    shift 2;;
      --record-ok-links )   RECORD_OK_LINKS=1;                  shift;;
      --suggest-snapshots ) SUGGEST_SNAPSHOTS=1;                shift;;
      --take-screenshots )  TAKE_PAGE_SHOT=1; CHROME_PATH="$2"; shift 2;;
      --start-url )         URL_START=$2;                       shift 2;;
      --end-url )           URL_LIMIT=$2;                       shift 2;;
      --upload )            UPLOAD_INFO=$2;                     shift 2;;
      * )                   echo "Invalid argument $1 detected. Aborting."; exit 1;;
  esac
done

# If the required arguments were not supplied, print help page and quit
if [ -z $LINKS_URL ] || [ -z $OUTPUT_DIR ]; then
   echo "Error: I did not receive one or both required arguments. Run me with the \"--help\" argument for documentation."
   exit 2
fi

# If user wants screenshots, make sure path to Chrome was passed in and is valid
if [ $TAKE_PAGE_SHOT -eq 1 ]; then
   if [ ! -f "$CHROME_PATH" ]; then
      echo "Error: You need to supply a path to the Google Chrome application in order to take screenshots."
      exit 3
   fi
fi

# Check that UPLOAD_INFO exists, if this argument was supplied
if [ ! -z $UPLOAD_INFO ] && [ ! -f "$UPLOAD_INFO" ]; then
   echo "Error: The file $UPLOAD_INFO supplied by the --upload argument does not appear to exist. Aborting."
   exit 4
fi

# Check that OUTPUT_DIR is a directory
if [ ! -d "$OUTPUT_DIR" ]; then
   echo "Error: The path $OUTPUT_DIR supplied by the --output argument does not appear to be a directory. Aborting."
   exit 5
fi

# Make timestamped folder inside OUTPUT_DIR for this session's log and screenshots
SAFE_TIME=$(date "+%Y-%m-%d--%H-%M-%S") # for use in path names
NICE_TIME=$(TZ=":UTC" date "+%a, %d %b %Y %k:%M:%S GMT") # matches format of LINKS_DATE, which is extracted from the extlinks file
OUTPUT_FOLDER="ValExtLinks ($SAFE_TIME)"
OUTPUT_PATH="$OUTPUT_DIR/$OUTPUT_FOLDER"
SHOT_PATH="$OUTPUT_PATH/Screenshots"
LOG_NAME="ValExtLinks report"
LOG_TXT="$OUTPUT_PATH/$LOG_NAME.txt"
LOG_RTF="$OUTPUT_PATH/$LOG_NAME.rtf"
LOG_HTM="$OUTPUT_PATH/$LOG_NAME.htm"
mkdir "$OUTPUT_PATH"
if [ $TAKE_PAGE_SHOT -eq 1 ]; then
   mkdir "$SHOT_PATH"
fi

# Check that 'mkdir' succeeded
if [ ! -d "$OUTPUT_PATH" ]; then
   echo "Error: I could not create the folder \"$OUTPUT_FOLDER\" inside the directory $OUTPUT_PATH. Aborting."
   exit 6
fi

# Get date on the file at LINKS_URL and print to log
LINKS_DATE=$(curl --silent --head $LINKS_URL | grep "Last-Modified")
if [ -z "$LINKS_DATE" ]; then
   echo "Error: I could not find the external links file at the path \"$LINKS_URL\" supplied by the --links argument. Aborting."
   exit 7
fi
LINKS_DATE=${LINKS_DATE#Last-Modified: }


### UTILITY FUNCTIONS ###
# Writes a plain-text header to TXT log file
function printTXTheader()
{
   valPrint t "Validate External Links report"
   valPrint t "generated $NICE_TIME"
   valPrint t "from data of $LINKS_DATE"
   valPrint t "script by Iritscen (contact: $MY_WIKI_PAGE)"
   valPrint t ""
}

# Writes the RTF header to RTF log file, then a centered title, then sets text to left-justified
function printRTFheader()
{
   valPrint r "{\rtf1\ansi\ansicpg1252\cocoartf1504\cocoasubrtf820
{\fonttbl\f0\fswiss\fcharset0 Helvetica;}
{\colortbl;\red255\green255\blue255;}
{\*\expandedcolortbl;;}
\margl1440\margr1440\vieww12600\viewh12100\viewkind0
\pard\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx6480\tx7200\tx7920\tx8640\li360\pardirnatural\qc\partightenfactor0

\f0\fs28 \cf0 \b1 Validate External Links report \b0\\
generated $NICE_TIME\\
from data of $LINKS_DATE\\
script by Iritscen ({\field{\*\fldinst{HYPERLINK \"$MY_WIKI_PAGE\"}}{\fldrslt contact}})
\\
\pard\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx6480\tx7200\tx7920\tx8640\li360\pardirnatural\partightenfactor0
\cf0 "
}

# Closes the RTF markup of the RTF log file
function printRTFfooter()
{
   valPrint r "}"
}

# Writes the HTML header to HTML log file
function printHTMheader()
{
   valPrint h "<html>
<head>
<title>Validate External Links report</title>
</head>
<body>
<h2>Validate External Links report</h2>
<h3>generated $NICE_TIME<br />
from data of $LINKS_DATE<br />
script by Iritscen (<a href=\"$MY_WIKI_PAGE\" target=\"_blank\">contact</a>)</h3>"
}

# Closes the HTML markup of the HTML log file
function printHTMfooter()
{
   valPrint h "</body>
</html>"
}

# The central logging function. The first parameter is a string composed of one or more characters that
# indicate which output to use: 'c' means console, 't' means the TXT log, 'r' means the RTF log, and
# 'h' means the HTML log. 'n' means "Don't print a newline at the end of the line." 'w' means "Don't
# pass console output through 'fmt'" ("fmt" fits the output to an 80-column CLI but can break special
# formatting and the 'n' option).
function valPrint()
{
   if [[ "$1" == *c* ]]; then
      if [[ "$1" == *n* ]]; then
         echo -n "$2"
      elif [[ "$1" == *w* ]]; then
         echo "$2"
      else
         echo "$2" | fmt -w 80
      fi
   fi
   if [[ "$1" == *t* ]]; then
      if [[ "$1" == *n* ]]; then
         echo -n "$2" >> "$LOG_TXT"
      else
         echo "$2" >> "$LOG_TXT"
      fi
   fi
   if [[ "$1" == *r* ]]; then
      if [[ "$1" == *n* ]]; then
         echo "$2" >> "$LOG_RTF"
      else
         echo "$2\\" >> "$LOG_RTF"
      fi
   fi
   if [[ "$1" == *h* ]]; then
      if [[ "$1" == *n* ]]; then
         echo "$2" >> "$LOG_HTM"
      else
         echo "$2<br />" >> "$LOG_HTM"
      fi
   fi
}

# Pluralize the string in parameter 1 if the number in parameter 2 is not 1
function pluralCheckNoun()
{
   if [ $2 -ne 1 ]; then
      if [[ $1 =~ x$ ]]; then
         echo $1es
      else
         echo $1s
      fi
   else
      echo $1
   fi
}

# Output "is" if parameter 1 is 1, otherwise "are"
function pluralCheckIs()
{
   if [ $1 -ne 1 ]; then
      echo "are"
   else
      echo "is"
   fi
}

# Output "was" if parameter 1 is 1, otherwise "were"
function pluralCheckWas()
{
   if [ $1 -ne 1 ]; then
      echo "were"
   else
      echo "was"
   fi
}

# Output "a " if parameter 1 is 1, otherwise nothing
function pluralCheckA()
{
   if [ $1 -eq 1 ]; then
      echo "a "
   fi
}

# Output "an " if parameter 1 is 1, otherwise nothing
function pluralCheckAn()
{
   if [ $1 -eq 1 ]; then
      echo "an "
   fi
}

# Upload HTML report using info specified in the --upload argument. ONLY USE "valPrint c" here, as the
# reports being saved to disk have already been closed.
function uploadReport()
{
   valPrint c "Uploading HTML report..."

   SCRIPT_PATH="$THIS_DIR/$EXPECT_SCRIPT_NAME"
   SFTP_USER_NAME_MARKER="user:"
   SFTP_PASSWORD_MARKER="pw:"
   SFTP_PORT_MARKER="port:"
   SFTP_PATH_MARKER="path:"
   SFTP_USER_NAME=$(grep $SFTP_USER_NAME_MARKER $UPLOAD_INFO)
   SFTP_USER_NAME=${SFTP_USER_NAME#$SFTP_USER_NAME_MARKER}
   SFTP_PASSWORD=$(grep $SFTP_PASSWORD_MARKER $UPLOAD_INFO)
   SFTP_PASSWORD=${SFTP_PASSWORD#$SFTP_PASSWORD_MARKER}
   SFTP_PORT=$(grep $SFTP_PORT_MARKER $UPLOAD_INFO)
   SFTP_PORT=${SFTP_PORT#$SFTP_PORT_MARKER}
   SFTP_PATH=$(grep $SFTP_PATH_MARKER $UPLOAD_INFO)
   SFTP_PATH=${SFTP_PATH#$SFTP_PATH_MARKER}

   expect "$SCRIPT_PATH" "$LOG_HTM" $SFTP_USER_NAME $SFTP_PASSWORD $SFTP_PORT $SFTP_PATH "$LOG_NAME.htm"

   valPrint c "Report was uploaded, unless an error message appears above."
}

# Prints session summary when script is done
function wrapupAndExit()
{
   # Get off progress line on console, drop down a line from last link in log, and close HTML table
   valPrint ctr ""
   valPrint h "</table><br />"

   # If we didn't finish processing the last URL, then the iterator is one too high
   if [ $FINISHED_LIST != "yes" ]; then
      let LINK_NUM-=1
      if [ $FINISHED_LIST == "no" ]; then
         valPrint ctrh "The session was canceled by the user."
      fi
   fi

   # Generate string with elapsed time
   END_RUN=$(date +%s)
   ELAPSED=$(echo $(($END_RUN - $START_RUN)) | awk '{printf "%d min. %d sec. elapsed", int($1/60), int($1%60)}')

   # Output results of session and close the log file's markup
   LINKS_PROCESSED=$((LINK_NUM-URL_START+1))
   LINKS_SKIPPED=$((SKIP_UNK_NS+SKIP_JS_PAGE+SKIP_BAD_URL+SKIP_NON_ASCII+SKIP_UNK_SUFFIX+SKIP_UNK_CODE))
   LINKS_CHECKED=$((LINKS_PROCESSED-LINKS_SKIPPED))
   valPrint ct "Summary ($ELAPSED):"
   valPrint r "\b1 Summary \b0 ($ELAPSED)"
   valPrint hn "<h3><span id=\"summary\">Summary ($ELAPSED)</span></h3>"
   valPrint ctrh "I finished processing $LINKS_PROCESSED of $LINK_COUNT $(pluralCheckNoun link $LINK_COUNT)."
   valPrint ctrh "I skipped $LINKS_SKIPPED $(pluralCheckNoun link $LINKS_SKIPPED), and found $FILE_LINKS $(pluralCheckNoun file $FILE_LINKS) and $PAGE_LINKS $(pluralCheckNoun page $PAGE_LINKS)."
   if [ $LINKS_SKIPPED -gt 0 ]; then valPrint ctrh "Skip breakdown: "; fi
   if [ $SKIP_UNK_NS -gt 0 ]; then valPrint ctrh "- $SKIP_UNK_NS unknown $(pluralCheckNoun namespace $SKIP_UNK_NS)"; fi
   if [ $SKIP_JS_PAGE -gt 0 ]; then valPrint ctrh "- $SKIP_JS_PAGE $(pluralCheckNoun link $SKIP_JS_PAGE) on $(pluralCheckA $SKIP_JS_PAGE)JavaScript $(pluralCheckNoun page $SKIP_JS_PAGE)"; fi
   if [ $SKIP_BAD_URL -gt 0 ]; then valPrint ctrh "- $SKIP_BAD_URL illegal $(pluralCheckNoun URL $SKIP_BAD_URL)"; fi
   if [ $SKIP_NON_ASCII -gt 0 ]; then valPrint ctrh "- $SKIP_NON_ASCII non-ASCII $(pluralCheckNoun URL $SKIP_NON_ASCII)"; fi
   if [ $SKIP_UNK_SUFFIX -gt 0 ]; then valPrint ctrh "- $SKIP_UNK_SUFFIX unknown URL $(pluralCheckNoun suffix $SKIP_UNK_SUFFIX)"; fi
   if [ $SKIP_UNK_CODE -gt 0 ]; then valPrint ctrh "- $SKIP_UNK_CODE unknown response $(pluralCheckNoun code $SKIP_UNK_CODE)"; fi
   valPrint ctrh "Out of the $LINKS_CHECKED links checked, $EI_LINKS could be $(pluralCheckAn $EI_LINKS)intrawiki $(pluralCheckNoun link $EI_LINKS), $IW_LINKS could be $(pluralCheckAn $IW_LINKS)interwiki $(pluralCheckNoun link $IW_LINKS), $OK_LINKS $(pluralCheckWas $OK_LINKS) OK, $RD_LINKS $(pluralCheckWas $RD_LINKS) $(pluralCheckA $RD_LINKS)redirection $(pluralCheckNoun notice $RD_LINKS), and $NG_LINKS $(pluralCheckWas $NG_LINKS) NG."
   if [ $SKIP_EXPECT_NG -gt 0 ]; then
      valPrint ctrh "$SKIP_EXPECT_NG/$NG_LINKS NG $(pluralCheckNoun link $NG_LINKS) went unlisted due to being found in the exceptions file."
   fi
   if [ $SKIP_EXPECT_EI -gt 0 ]; then
      valPrint ctrh "$SKIP_EXPECT_EI/$EI_LINKS external internal $(pluralCheckNoun link $EI_LINKS) went unlisted due to being found in the exceptions file."
   fi
   if [ $SKIP_EXPECT_IW -gt 0 ]; then
      valPrint ctrh "$SKIP_EXPECT_IW/$IW_LINKS potential intrawiki $(pluralCheckNoun link $IW_LINKS) went unlisted due to being found in the exceptions file."
   fi
   valPrint trh "ValExtLinks says goodbye."
   printRTFfooter
   printHTMfooter

   # Upload report if this was requested
   if [ ! -z $UPLOAD_INFO ] && [ $FINISHED_LIST != "no" ]; then
      uploadReport
   fi

   # Really quit now
   valPrint c "ValExtLinks says goodbye."
   exit 0
}
trap wrapupAndExit INT


### INITIALIZATION ###
# Print opening message to console and log files
valPrint cw "$(tput bold)--Validate External Links--$(tput sgr0)"
printTXTheader
printRTFheader
printHTMheader

# Attempt to download file at LINKS_URL, then check that it succeeded
valPrint cwtrh "Downloading list of external links from $LINKS_URL."
LINKS_FILE_NAME=$(echo "$LINKS_URL" | sed 's/.*\///')
LINKS_FILE="$OUTPUT_PATH/$LINKS_FILE_NAME"
curl --silent -o "$LINKS_FILE" $LINKS_URL
if [ ! -f "$LINKS_FILE" ]; then
   echo "The download of $LINKS_URL appears to have failed. Aborting."
   wrapupAndExit
fi

# Attempt to download file at EXCEPT_URL, then check that it succeeded
if [ ! -z $EXCEPT_URL ]; then
   valPrint cwtrh "Downloading list of reporting exceptions from $EXCEPT_URL."
   EXCEPT_FILE_NAME=$(echo "$EXCEPT_URL" | sed 's/.*\///')
   EXCEPT_FILE="$OUTPUT_PATH/$EXCEPT_FILE_NAME"
   curl --silent -o "$EXCEPT_FILE" $EXCEPT_URL
   if [ ! -f "$EXCEPT_FILE" ]; then
      echo "The download of $EXCEPT_URL appears to have failed. Aborting."
      wrapupAndExit
   fi
fi

# Pipe 'cat' to 'wc' because passing LINKS_FILE to 'wc' would print the file name after the line count
LINK_COUNT_STRING=$(cat "$LINKS_FILE" | wc -l)

# Number of URLs is number of lines minus one (first line is column header row for the CSV)
LINK_COUNT=$(echo "${LINK_COUNT_STRING}" | tr -d '[:space:]')
let LINK_COUNT-=1

# Calculate number of URLs to consider
if [ $URL_LIMIT -ne 0 ] && [ $URL_LIMIT -lt $LINK_COUNT ]; then
   valPrint ctrh "Found $LINK_COUNT links to process. I will be considering $((URL_LIMIT-URL_START+1)) of them, from link $URL_START to $URL_LIMIT."
elif [ $URL_START -ne 1 ]; then
   valPrint ctrh "Found $LINK_COUNT links to process. I will be considering $((LINK_COUNT-URL_START+1)) of them, from link $URL_START to $LINK_COUNT."
else
   valPrint ctrh "Found $LINK_COUNT links to process. I will be considering all of them."
fi

# Print settings to console and log
declare -a SETTINGS_MSG=(I will be checking the response code of each link "and will" take a screenshot of each page. Pages that are OK will "also" be logged. I "will" ask the Internet Archive for a suggested snapshot URL for each NG page. "I will not report links that are listed in the exceptions file.")
if [ $TAKE_PAGE_SHOT -eq 0 ]; then SETTINGS_MSG[10]="but will not"; fi
if [ $RECORD_OK_LINKS -eq 0 ]; then SETTINGS_MSG[22]="not"; fi
if [ $SUGGEST_SNAPSHOTS -eq 0 ]; then SETTINGS_MSG[26]="will not"; fi
if [ -z $EXCEPT_URL ] || [ $RECORD_OK_LINKS -eq 1 ]; then SETTINGS_MSG[40]=""; fi
SETTINGS_STR=${SETTINGS_MSG[@]}
valPrint ctrh "$SETTINGS_STR"
valPrint tr "A summary of my findings will be found at the bottom of the report."
valPrint h "A summary of my findings will be found at the <a href=\"#summary\">bottom</a> of the report."
valPrint trh ""

# Print legend to logs
valPrint t "Legend:"
valPrint r "\b1 Legend \b0"
valPrint hn "<h3>Legend</h3>"
valPrint trh "OK = URL seems to be working."
valPrint trh "NG = URL no longer seems to work. You should click each URL marked as NG before attempting to fix it, because false negatives will occur from time to time due to hiccups in the Internet. Please report any persistent false negatives or other issues to Iritscen. An NG link should be followed by a link to the Internet Archive's Wayback Machine which may help you repair the link. If the link cannot be repaired, you can disable it on the wiki (which prevents it from showing up in future ValExtLinks reports) by wrapping it in nowiki tags."
valPrint trh "RD = The server responding to this URL is saying that the page moved and you should instead use the supplied new URL. Some RD links represent minor adjustments in the organization of a web site, and some are soft 404s (the file/page has been removed and you are being redirected to something like the main page of the web site). You will have to look at the new URL yourself to determine if it represents an OK link and the link on the wiki should be updated to this one, or if the desired file/page is actually gone and we need to replace the wiki link with an Internet Archive snapshot link -- or disable the URL if it has not been archived."
valPrint trh "EI = URL is an external link to an internal page and should be converted to an intrawiki link using the suggested markup."
valPrint trh "IW = URL is an external link to a fellow wiki and should be converted to an interwiki link using the suggested markup."
valPrint t "(xxx) = Unix tool 'curl' obtained this HTTP response status code (see here for code reference: $HTTP_CODES)."
valPrint r "(xxx) = Unix tool 'curl' obtained this HTTP response status code (see {\field{\*\fldinst{HYPERLINK \"$HTTP_CODES\"}}{\fldrslt here}} for code reference)."
valPrint h "(xxx) = Unix tool 'curl' obtained this HTTP response status code (see <a href=\"$HTTP_CODES\" target=\"_blank\">here</a> for code reference)."
valPrint t "(000-xx) = 'curl' did not get an HTTP response code, but returned this exit code (see here for code reference: $CURL_CODES)."
valPrint r "(000-xx) = 'curl' did not get an HTTP response code, but returned this exit code (see {\field{\*\fldinst{HYPERLINK \"$CURL_CODES\"}}{\fldrslt here}} for code reference)."
valPrint h "(000-xx) = 'curl' did not get an HTTP response code, but returned this exit code (see <a href=\"$CURL_CODES\" target=\"_blank\">here</a> for code reference)."
valPrint trh "IA suggests = Last available snapshot suggested by the Internet Archive."
valPrint trh "Try browsing = The Archive occasionally fails to return a snapshot URL even when one exists, so you will need to check for a snapshot manually using this link to the Wayback Machine before concluding that a site has not been archived."
valPrint trh ""


### MAIN LOOP ###
START_RUN=$(date +%s)
# Process each line of the .csv in LINKS_FILE
for LINE in `cat "$LINKS_FILE"`; do
   let LINK_NUM+=1

   # First line is the column header row for the CSV, so let's verify that the format hasn't changed
   if [ $SKIPPED_HEADER_ROW -eq 0 ]; then
      if [ $LINE == "namespace,title,target" ]; then
         SKIPPED_HEADER_ROW=1
         LINK_NUM=0 # this line is it's not a link, so reset the link counter
         valPrint hn "<table>"
         continue
      else
         valPrint ctrh "Did not find expected header row in $LINKS_FILE. Aborting."
         wrapupAndExit
      fi
   fi

   # Skip this link if we are not at URL_START yet
   if [ $LINK_NUM -lt $URL_START ]; then
      continue
   fi

   # Stop if we are at the limit declared for testing purposes
   if [ $URL_LIMIT -gt 0 ] && [ $LINK_NUM -gt $URL_LIMIT ]; then
      FINISHED_LIST="limit"
      wrapupAndExit
   fi

   # Print progress to screen
   if [ $LINK_NUM -gt 1 ]; then
      printf "\e[1A\n" # erase previous progress message so that new one appears in its place
   fi
   valPrint cn "Evaluating URL $LINK_NUM/$LINK_COUNT..."

   # The number of the namespace is the element before the first comma on the line
   NS_ID=${LINE%%,*}

   # Find namespace number in NS_IDS and use it to look up namespace's name in NS_NAMES
   NS_NAME=""
   a=0
   while [ "x${NS_IDS[$a]}" != "x" ]; do # once this evaluates to "x", the array is done
      if [ $NS_ID == "NULL" ]; then
         break
      elif [ $NS_ID -eq ${NS_IDS[$a]} ]; then
         NS_NAME="${NS_NAMES[$a]}"
         break
      fi
      let a+=1
   done
   if [ "$NS_NAME" == "" ]; then
      if [ $NS_ID == "NULL" ]; then
         valPrint tr "Skipping URL on line $LINK_NUM because the namespace (and probably the page too) is \"NULL\". Probably the link is no longer in existence on the wiki."
      else
         valPrint tr "Skipping URL on line $LINK_NUM found on page $PAGE_NAME because I could not find a name for namespace ID $NS_ID."
      fi
      let SKIP_UNK_NS+=1
      continue
   fi

   # The name of the page is everything between the namespace ID and the next comma on the line (commas
   # in page names will break this)
   PAGE_NAME=${LINE#$NS_ID,}
   PAGE_NAME=${PAGE_NAME%%,*}

   # We don't want to consider wiki pages ending in .js, as the parser cannot reliably isolate URLS in
   # JavaScript code, so it will return erroneous links
   PAGE_NAME_SUFFIX=$(echo $PAGE_NAME | sed 's/.*\.//')
   if [ $PAGE_NAME_SUFFIX == "js" ]; then
      valPrint tr "Skipping URL on line $LINK_NUM because it was found on JavaScript page $PAGE_NAME."
      let SKIP_JS_PAGE+=1
      continue
   fi

   # Build longer wiki page URLs from namespace and page names
   FULL_PAGE_PATH=http://$WIKI_PATH/$NS_NAME:$PAGE_NAME
   LOCAL_PAGE_PATH=$NS_NAME:$PAGE_NAME
   # Namespace "Main:" cannot be a part of the path; it's an implicit namespace, and naming it
   # explicitly breaks the link
   if [ $NS_ID -eq 0 ]; then
      FULL_PAGE_PATH=http://$WIKI_PATH/$PAGE_NAME
      LOCAL_PAGE_PATH=$PAGE_NAME
   fi

   # The URL being linked to is everything after the previous two fields (this allows commas to be in
   # the URLs, but a comma in the previous field, the page name, will break this)
   URL=${LINE#$NS_ID,$PAGE_NAME,}

   # Scan for illegal characters
   if [[ $URL == *[$ILLEGAL_CHARS]* ]]; then
      valPrint tr "Skipping URL $URL (found on page $PAGE_NAME) because it contains characters illegal in a URL."
      let SKIP_BAD_URL+=1
      continue
   fi

   # Now we need to know if the URL is for a file or a web page. First step is to determine if the
   # URL ends in a suffix
   HAS_SUFFIX=0

   # If the URL ends in something like ".php?foo=bar", strip everything from the '?' onward
   CLEAN_URL=${URL%%\?*}

   # If the URL ends in something like "#section_15", strip everything from the '#' onward
   CLEAN_URL=${CLEAN_URL%%\#*}

   # 'sed' cannot handle Unicode in my Bash shell, so skip this URL and make user check it
   if [[ $CLEAN_URL == *[![:ascii:]]* ]]; then
      valPrint tr "Skipping URL $URL (found on page $PAGE_NAME) because I cannot handle non-ASCII characters."
      let SKIP_NON_ASCII+=1
      continue
   fi

   # Isolate the characters after the last period and after the last slash
   POST_DOT=$(echo "$CLEAN_URL" | sed 's/.*\.//')
   POST_SLASH=$(echo "$CLEAN_URL" | sed 's/.*\///')

   # If the last period comes after the last slash, then the URL ends in a suffix
   POST_DOT_LENGTH=$(echo | awk -v input=$POST_DOT '{print length(input)}')
   POST_SLASH_LENGTH=$(echo | awk -v input=$POST_SLASH '{print length(input)}')
   if [ $POST_DOT_LENGTH -lt $POST_SLASH_LENGTH ]; then
      HAS_SUFFIX=1
   else
      HAS_SUFFIX=0
   fi

   # Now we need to know if the URL is for a file. If the URL ends in a suffix, compare it against the
   # known URL endings supplied in HTTP_FILES and HTTP_TLDS_AND_PAGES
   IS_FILE=-1
   if [ $HAS_SUFFIX -eq 0 ]; then
      IS_FILE=0
   else
      # Turn off case sensitivity while we compare suffixes
      shopt -s nocasematch

      # Special case: URLs ending in something like "/productID.304297400" are pages, not files, so if
      # the URL's suffix is all numbers, we are looking at the end of a web page URL
      if [[ $POST_DOT =~ ^-?[0-9]+$ ]]; then
         IS_FILE=0
      fi
      
      # If we did not identify this URL as a web page above, we need to compare the suffix against known
      # file extensions
      if [ $IS_FILE -eq -1 ]; then
         for EXTENSION in "${HTTP_FILES[@]}"; do
            if [[ $EXTENSION == $POST_DOT ]]; then # double brackets respect the nocasematch setting
               IS_FILE=1
               break
            fi
         done
      fi

      # If we did not identify this URL as a file above, we need to compare the suffix against known
      # pages and TLDs in order to rule out that this is an unknown file suffix or TLD that the user
      # needs to add to the HTTP_FILES or HTTP_TLDS_AND_PAGES array
      if [ $IS_FILE -eq -1 ]; then
         for PAGE_OR_TLD in "${HTTP_TLDS_AND_PAGES[@]}"; do
            if [[ $PAGE_OR_TLD == $POST_DOT ]]; then
               IS_FILE=0
               break
            fi
         done
      fi

      # Turn case sensitivity back on in Bash
      shopt -u nocasematch
   fi

   # If this suffix escaped identification as either a file, page or TLD, inform the user
   STR_TYPE=""
   if [ $IS_FILE -eq -1 ]; then
      valPrint tr "Skipping URL $URL (found on page $PAGE_NAME) because I encountered the unknown URL ending $POST_DOT. Please add this ending to the appropriate array in this script, HTTP_FILES or HTTP_TLDS_AND_PAGES."
      let SKIP_UNK_SUFFIX+=1
      continue
   elif [ $IS_FILE -eq 1 ]; then
      STR_TYPE="file"
      let FILE_LINKS+=1
   elif [ $IS_FILE -eq 0 ]; then
      STR_TYPE="page"
      let PAGE_LINKS+=1
   fi

   # Get response code using 'curl' to see if this link is valid; the --insecure option avoids an
   # issue with sites that require HTTPS
   CURL_CODE=$(curl -o /dev/null --silent --insecure --head --user-agent '"$AGENT"' --max-time 10 --write-out '%{http_code}\n' $URL)
   CURL_ERR=$(echo $?)
   CURL_RESULT=$CURL_CODE

   # Tack 'curl' exit code onto result string if 'curl' returned "000" (no HTTP response)
   if [ $CURL_CODE == "000" ]; then
      CURL_RESULT="$CURL_RESULT-$CURL_ERR"
   fi

   # Begin to determine our status code for this URL (EI, IW, OK, RD, or NG)
   STATUS="??"
   NEW_URL=""
   INTERWIKI_INDEX=-1

   # First make sure that this isn't an "external internal" link to our own wiki that can be replaced
   # by "[[page_name]]". If it uses a special access URL beginning with "/w/", let it pass, as it
   # probably cannot be replaced by "[[ ]]" markup
   if [[ $URL == *$WIKI_PATH* ]] && [[ $URL != *$WIKI_PATH/w/* ]]; then
      STATUS="EI"
      let EI_LINKS+=1
   fi

   # If it's not, check if this is a link to a domain that we have an interwiki prefix for
   if [ $STATUS == "??" ]; then
      for ((i = 0; i < ${#INTERWIKI_DOMAINS[@]}; ++i)); do
         if [[ $URL == *${INTERWIKI_DOMAINS[$i]}* ]] && [[ $URL != *${INTERWIKI_DOMAINS[$i]}/w/* ]]; then
            STATUS="IW"
            let IW_LINKS+=1
            INTERWIKI_INDEX=$i
            break
         fi
      done
   fi

   # If we didn't match an interwiki domain, see if the status code is in our "OK" codes list
   if [ $STATUS == "??" ]; then
      for CODE in "${OK_CODES[@]}"; do
         if [[ $CODE == $CURL_CODE ]]; then
            STATUS="OK"
            let OK_LINKS+=1
            break
         fi
      done
   fi

   # If we didn't get a match with the "OK" codes, check it against the "RD" codes
   if [ $STATUS == "??" ]; then
      for CODE in "${RD_CODES[@]}"; do
         if [[ $CODE == $CURL_CODE ]]; then
            # Get URL header again in order to retrieve the URL we are being redirected to
            NEW_URL=$(curl -o /dev/null --silent --insecure --head --user-agent '"$AGENT"' --max-time 10 --write-out '%{redirect_url}\n' $URL)

            # Filter out cases where the redirect URL is just the original URL with https:// instead of
            # http://, or with an added '/' at the end. These corrections happen a lot and are not
            # important to us.
            URL_NO_PROTOCOL=${URL#http://}
            URL_NO_PROTOCOL=${URL_NO_PROTOCOL%/}
            NEW_URL_NO_PROTOCOL=${NEW_URL#https://}
            NEW_URL_NO_PROTOCOL=${NEW_URL_NO_PROTOCOL%/}

            # Sometimes 'curl' fails to get the redirect_url due to time-out or bad web site config
            NEW_URL_LENGTH=$(echo | awk -v input=$NEW_URL_NO_PROTOCOL '{print length(input)}')
            if [ $NEW_URL_LENGTH -lt $MIN_URL_LENGTH ]; then
               NEW_URL_NO_PROTOCOL="[new URL not retrieved]"
            fi

            # If the URLs match after the above filters were applied, then the link is OK
            if [ $URL_NO_PROTOCOL == $NEW_URL_NO_PROTOCOL ]; then
               STATUS="OK"
               let OK_LINKS+=1
            else
               STATUS="RD"
               let RD_LINKS+=1
            fi
            break
         fi
      done
   fi

   # If we didn't get a match with the "RD" codes, check it against the "NG" codes
   if [ $STATUS == "??" ]; then
      for CODE in "${NG_CODES[@]}"; do
         if [[ $CODE == $CURL_CODE ]]; then
            STATUS="NG"
            let NG_LINKS+=1
            break
         fi
      done
   fi

   # If we didn't match a known status code, advise the reader
   if [ $STATUS == "??" ]; then
      valPrint tr "Skipping URL $URL (found on page $PAGE_NAME) because I encountered the unknown return code $CURL_CODE."
      let SKIP_UNK_CODE+=1
      continue
   fi

   # Check problem links against exceptions file before proceeding
   if [ $STATUS != "OK" ] && [ ! -z $EXCEPT_URL ]; then
      # The code we expect to find in the exceptions file is either the 'curl' result or "EI"/"IW"
      EXPECT_CODE="$CURL_RESULT"
      if [ $STATUS == "EI" ]; then
         EXPECT_CODE="EI"
      elif [ $STATUS == "IW" ]; then
         EXPECT_CODE="IW"
      fi

      # Look for link in exceptions file and make sure its listed result code and wiki page also match
      GREP_RESULT=$(grep --max-count=1 "$URL" "$EXCEPT_FILE")
      EXCEPT_PAGE=${GREP_RESULT##*,}
      if [ "$EXCEPT_PAGE" == "*" ] || [ "$EXCEPT_PAGE" == $LOCAL_PAGE_PATH ]; then
         EXCEPT_CODE=${GREP_RESULT%%,*}
         if [ "$EXCEPT_CODE" == "$EXPECT_CODE" ]; then
            valPrint tr "Skipping URL $URL (found on page $PAGE_NAME) because its expected result, $EXPECT_CODE, is listed in the exceptions file."
            if [ $STATUS == "EI" ]; then
               let SKIP_EXPECT_EI+=1
            elif [ $STATUS == "IW" ]; then
               let SKIP_EXPECT_IW+=1
            else
               let SKIP_EXPECT_NG+=1
            fi
            continue
         fi
      fi
   fi

   # If appropriate, record this link to the log, with clickable URLs when possible
   if [ $STATUS != "OK" ] || [ $RECORD_OK_LINKS -eq 1 ]; then
      # Stupid hack since the strings "IW" and "EI" are narrower than "OK", "RD", or "NG" and it takes
      # an extra tab to get to the desired level of indentation in the RTF log
      RTF_TABS="        "
      if [ $STATUS == "IW" ] || [ $STATUS == "EI" ]; then
         RTF_TABS="             "
      fi
      
      # Record link and its wiki page in TXT, RTF, and HTML markup
      valPrint t "$STATUS ($CURL_RESULT) $STR_TYPE $URL"
      valPrint t "  linked from $FULL_PAGE_PATH"
      valPrint r "$STATUS ($CURL_RESULT)$RTF_TABS$STR_TYPE      {\field{\*\fldinst{HYPERLINK \"$URL\"}}{\fldrslt $URL}}"
      valPrint r "              linked from     {\field{\*\fldinst{HYPERLINK \"$FULL_PAGE_PATH\"}}{\fldrslt $LOCAL_PAGE_PATH}}"
      valPrint hn "<tr><td style=\"white-space:nowrap\">$STATUS ($CURL_RESULT)</td><td align=\"right\">$STR_TYPE</td><td><a href=\"$URL\" target=\"_blank\">$URL</a></td></tr>"
      valPrint hn "<tr><td colspan=\"2\" align=\"right\">linked from</td><td><a href=\"$FULL_PAGE_PATH\" target=\"_blank\">$LOCAL_PAGE_PATH</a></td></tr>"

      # Record redirect URL if one was given by a 3xx response page
      if [ $STATUS == "RD" ]; then
         valPrint t "  Server suggests $NEW_URL"
         valPrint r "   Server suggests {\field{\*\fldinst{HYPERLINK \"$NEW_URL\"}}{\fldrslt $NEW_URL}}"
         valPrint hn "<tr><td colspan=\"2\" align=\"right\">Server suggests</td><td><a href=\"$NEW_URL\" target=\"_blank\">$NEW_URL</a></td></tr>"
      fi

      # Notify reader if we can use an intrawiki link for this URL
      if [ $STATUS == "EI" ]; then
         INTRA_PAGE=${URL#*://*/}
         valPrint t "  Just use [[$INTRA_PAGE]]"
         valPrint r "           Just use [[$INTRA_PAGE]]"
         valPrint hn "<tr><td colspan=\"2\" align=\"right\">Just use</td><td>[[$INTRA_PAGE]]</td></tr>"
      fi

      # Notify reader if we can use an interwiki prefix for this URL
      if [ $STATUS == "IW" ]; then
         INTER_PAGE=$(echo "$URL" | sed 's/.*\///')
         valPrint t "  You can use [[${INTERWIKI_PREFIXES[$INTERWIKI_INDEX]}:$INTER_PAGE]]"
         valPrint r "           You can use [[${INTERWIKI_PREFIXES[$INTERWIKI_INDEX]}:$INTER_PAGE]]"
         valPrint hn "<tr><td colspan=\"2\" align=\"right\">You can use</td><td>[[${INTERWIKI_PREFIXES[$INTERWIKI_INDEX]}:$INTER_PAGE]]</td></tr>"
      fi

      # Query Internet Archive for latest "OK" snapshot for "NG" page
      if [ $STATUS == "NG" ] && [ $SUGGEST_SNAPSHOTS -eq 1 ]; then
         ARCHIVE_QUERY=$(curl --silent --max-time 10 "$ARCHIVE_API?url=$URL&$ARCHIVE_OK_CODES")

         # If a "closest" snapshot was received...
         if [[ "$ARCHIVE_QUERY" == *\"closest\":* ]]; then
            # In case the URL has a shebang in it (like mega.nz links do), escape the '!' to break it
            ARCHIVE_QUERY=$(echo "$ARCHIVE_QUERY" | sed 's/#!/#\\!/')

            # ...isolate "url" property in the response that follows the "closest" tag
            SNAPSHOT_URL=${ARCHIVE_QUERY##*\"closest\":} # everything after '"closest":'
            SNAPSHOT_URL=${SNAPSHOT_URL##*\"url\": \"} # everything after '"url": "'
            SNAPSHOT_URL=${SNAPSHOT_URL%%\"*} # everything before '"'

            # Inform the user of the snapshot URL
            valPrint t "  IA suggests $SNAPSHOT_URL"
            valPrint r "                IA suggests {\field{\*\fldinst{HYPERLINK \"$SNAPSHOT_URL\"}}{\fldrslt $SNAPSHOT_URL}}"
            valPrint hn "<tr><td colspan=\"2\" align=\"right\">IA suggests</td><td><a href=\"$SNAPSHOT_URL\" target=\"_blank\">$SNAPSHOT_URL</a></td></tr>"
         else # ...otherwise give generic Wayback Machine link for this URL
            valPrint t " Try browsing $ARCHIVE_GENERIC/$URL"
            valPrint r "                Try browsing {\field{\*\fldinst{HYPERLINK \"$ARCHIVE_GENERIC/$URL\"}}{\fldrslt $ARCHIVE_GENERIC/$URL}}"
            valPrint hn "<tr><td colspan=\"2\" align=\"right\">Try browsing</td><td><a href=\"$ARCHIVE_GENERIC/$URL\" target=\"_blank\">$ARCHIVE_GENERIC/$URL</a></td></tr>"
         fi
      fi
   fi
   
   # If this is a page and it seems to be there, take screenshot of it using headless Google Chrome
   if [ $IS_FILE -eq 0 ] && [ $TAKE_PAGE_SHOT -eq 1 ] && [ $STATUS == "OK" ]; then
      # Sanitize URL by removing "http(s)://" and converting any other colons or slashes to underscores
      SHOT_NAME=$(echo "$URL" | sed 's/https*\:\/\///' | sed 'y/:\//__/')
      SHOT_FILE="$SHOT_PATH/$SHOT_NAME.png"

      # Don't take screenshot if we already encountered this page and screenshotted it
      if [ ! -f "$SHOT_FILE" ]; then
         "$CHROME_PATH" --headless --disable-gpu --screenshot --window-size=1500,900 $URL > /dev/null 2>&1
         if [ -f "$WORKING_DIR/$CHROME_SCREENSHOT" ]; then
            mv -n "$WORKING_DIR/$CHROME_SCREENSHOT" "$SHOT_FILE"
         else
            valPrint trh "Screenshot of URL $URL seems to have failed!"
         fi
      else
         valPrint trh "Skipping screenshot of URL $URL because $SHOT_FILE already exists."
      fi
   fi
done
FINISHED_LIST="yes"
wrapupAndExit