--- Validate External Links/validate_external_links.sh	2023/05/16 01:10:09	1183
+++ Validate External Links/validate_external_links.sh	2023/05/21 22:22:55	1184
@@ -33,6 +33,7 @@ LINKS_URL=""           # download extern
 EXCEPT_URL=""          # location of wiki page with a list of exceptions for 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
+ONLY_200_OK=0          # only treat code 200 as "OK" and not any other code in OK_CODES
 SHOW_SLASH=0           # record issue when a slash is added to the end of a URL
 SHOW_HTTPS=0           # record issue when "http" is upgraded to "https"
 SHOW_YT_RD=0           # record redirection for a youtu.be URL expanding to the full URL
@@ -67,7 +68,8 @@ declare -a NS_IDS=(-2 -1 0 1 2 3 4 5 6 7
 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.
+# This determines whether the script tries to take a screenshot of the URL (when screenshots are
+# requested).
 declare -a HTTP_FILES=(3ds 7z avi BINA blend bsl cpp csv dae dll dmg doc east exe fbx first63 flv gamedata gif jpg last32 log m4a mhm mov mp3 mp4 oni ONWC pdf png psd py rar tga TRMA txt vbs wav wmv xaf xcf xlsx xml zip)
 declare -a HTTP_TLDS_AND_PAGES=(abstract action ars asp aspx cfm cgi com css de do full htm html it js jsp net org pgi php php3 phtml pl ru shtml stm uk x)
 
@@ -134,10 +136,10 @@ NAME
 SYNOPSIS
        validate_external_links.sh --help
        validate_external_links.sh --links URL --output DIR [--exceptions URL]
-          [--record-ok-links] [--show-added-slashes] [--show-https-upgrades]
-          [--show-yt-redirects] [--suggest-snapshots] [--check-archive-links]
-          [--take-screenshots FILE] [--timeout NUM] [--start-url NUM]
-          [--end-url NUM] [--upload FILE]
+          [--record-ok-links] [--only-200-ok] [--show-added-slashes]
+          [--show-https-upgrades] [--show-yt-redirects] [--suggest-snapshots]
+          [--check-archive-links] [--take-screenshots FILE] [--timeout NUM]
+          [--start-url NUM] [--end-url NUM] [--upload FILE]
 
 DESCRIPTION
        This script parses a list of external links found in the OniGalore wiki
@@ -172,6 +174,10 @@ OPTIONS
                                beginning with "file://".
        --record-ok-links       Log a link in the report even if its response
                                code is "OK".
+       --only-200-ok           Only treat response code 200 as "OK". Normally
+                               several additional codes are treated as "OK" (see
+                               the array OK_CODES in script) because they are
+                               typically not an indicator of a bad link.
        --show-added-slashes    Report on redirects that simply add a '/' to the
                                end of the URL.
        --show-https-upgrades   Report on redirects that simply upgrade a
@@ -225,6 +231,7 @@ while (( "$#" )); do
       --exceptions )           EXCEPT_URL="$2";                    shift 2;;
       --output )               OUTPUT_DIR="$2";                    shift 2;;
       --record-ok-links )      RECORD_OK_LINKS=1;                  shift;;
+      --only-200-ok )          ONLY_200_OK=1;                      shift;;
       --show-added-slashes )   SHOW_SLASH=1;                       shift;;
       --show-https-upgrades )  SHOW_HTTPS=1;                       shift;;
       --show-yt-redirects )    SHOW_YT_RD=1;                       shift;;
@@ -704,6 +711,9 @@ valPrint ctrh "Site query timeout: $TIME
 valPrint ctrhn "Show OK links: "
 if [ $RECORD_OK_LINKS -eq 1 ]; then valPrint ctrh "Yes"; else valPrint ctrh "No"; fi
 
+valPrint ctrhn "Treat these response codes as OK: "
+if [ $ONLY_200_OK -eq 1 ]; then valPrint ctrh "200"; else valPrint ctrh "${OK_CODES[*]}"; fi
+
 valPrint ctrhn "Take screenshots: "
 if [ $TAKE_PAGE_SHOT -eq 1 ]; then valPrint ctrh "Yes"; else valPrint ctrh "No"; fi
 
@@ -1006,6 +1016,10 @@ for LINE in `cat "$LINKS_FILE"`; do
    # 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 [ $ONLY_200_OK -eq 1 ] && [ $CODE -ne 200 ]; then
+            continue
+         fi
+      
          if [[ $CODE == $CURL_CODE ]]; then
             STATUS="OK"
             let OK_LINKS+=1
@@ -1142,6 +1156,19 @@ for LINE in `cat "$LINKS_FILE"`; do
             break
          fi
       done
+      # Also check it against the "OK" codes besides 200 if the --only-200-ok argument was received
+      if [ $ONLY_200_OK -eq 1 ]; then
+         for CODE in "${OK_CODES[@]}"; do
+            if [ $CODE -eq 200 ]; then
+               continue
+            fi
+            if [[ $CODE == $CURL_CODE ]]; then
+               STATUS="NG"
+               let NG_LINKS+=1
+               break
+            fi
+         done
+      fi
    fi
 
    # If we didn't match a known status code, advise the reader