sysutils/gh-md-toc: Update to 0.10.0

ChangeLog: https://github.com/ekalinin/github-markdown-toc/releases/tag/0.10.0

* Update for a new GH HTML layout

PR:		276219

3 files changed, 211 insertions, 53 deletions

diff --git a/sysutils/gh-md-toc/Makefile b/sysutils/gh-md-toc/Makefile
index 6c4ba85a795d..aa18724b6d97 100644
--- a/sysutils/gh-md-toc/Makefile
+++ b/sysutils/gh-md-toc/Makefile
@@ -1,5 +1,5 @@
 PORTNAME=	gh-md-toc
-DISTVERSION=	0.8.0
+DISTVERSION=	0.10.0
 CATEGORIES=	sysutils
 
 MAINTAINER=	DtxdF@disroot.org
@@ -9,7 +9,8 @@ WWW=		https://github.com/ekalinin/github-markdown-toc
 LICENSE=	MIT
 LICENSE_FILE=	${WRKSRC}/LICENSE
 
-RUN_DEPENDS=	curl>0:ftp/curl
+RUN_DEPENDS=	curl>0:ftp/curl \
+		jq>0:textproc/jq
 
 USE_GITHUB=	yes
 GH_ACCOUNT=	ekalinin
@@ -21,7 +22,7 @@ PLIST_FILES=	bin/gh-md-toc
 
 PORTDOCS=	README.md
 
-OPTIONS_DEFINE=		DOCS
+OPTIONS_DEFINE=	DOCS
 
 do-install:
 	${INSTALL_SCRIPT} ${WRKSRC}/gh-md-toc ${STAGEDIR}${PREFIX}/bin
diff --git a/sysutils/gh-md-toc/distinfo b/sysutils/gh-md-toc/distinfo
index 16f7fc92da8d..1ed73e90898b 100644
--- a/sysutils/gh-md-toc/distinfo
+++ b/sysutils/gh-md-toc/distinfo
@@ -1,3 +1,3 @@
-TIMESTAMP = 1687043066
-SHA256 (ekalinin-github-markdown-toc-0.8.0_GH0.tar.gz) = 7f6079026a32c03d6e7254e5921b94f6802c2d52c9d74d85a1f8015d864b58d0
-SIZE (ekalinin-github-markdown-toc-0.8.0_GH0.tar.gz) = 10927
+TIMESTAMP = 1710461706
+SHA256 (ekalinin-github-markdown-toc-0.10.0_GH0.tar.gz) = d123d360b76e0ba97974cb32fb386905612d6b0ebb0a018161056062d9b44023
+SIZE (ekalinin-github-markdown-toc-0.10.0_GH0.tar.gz) = 24032
diff --git a/sysutils/gh-md-toc/files/patch-gh-md-toc b/sysutils/gh-md-toc/files/patch-gh-md-toc
index ac4e37045b34..c2c5213ec763 100644
--- a/sysutils/gh-md-toc/files/patch-gh-md-toc
+++ b/sysutils/gh-md-toc/files/patch-gh-md-toc
@@ -1,4 +1,4 @@
---- gh-md-toc.orig	2023-06-30 18:59:15 UTC
+--- gh-md-toc.orig	2024-03-03 10:54:38 UTC
 +++ gh-md-toc
 @@ -1,4 +1,4 @@
 -#!/usr/bin/env bash
@@ -6,22 +6,23 @@
  
  #
  # Steps:
-@@ -34,12 +34,10 @@ gh_user_agent="gh-md-toc v$gh_toc_version"
+@@ -34,12 +34,10 @@ gh_toc_load() {
  gh_toc_load() {
      local gh_url=$1
  
 -    if type curl &>/dev/null; then
-+    if type curl > /dev/null 2>&1; then
-         curl --user-agent "$gh_user_agent" -s "$gh_url"
+-        curl --user-agent "$gh_user_agent" -s "$gh_url"
 -    elif type wget &>/dev/null; then
 -        wget --user-agent="$gh_user_agent" -qO- "$gh_url"
++    if type curl > /dev/null 2>&1; then
++        curl --location --fail --user-agent "$gh_user_agent" -s "$gh_url"
      else
 -        echo "Please, install 'curl' or 'wget' and try again."
 +        echo "Please, install 'curl' and try again."
          exit 1
      fi
  }
-@@ -56,7 +54,7 @@ gh_toc_md2html() {
+@@ -58,7 +56,7 @@ gh_toc_md2html() {
      if [ ! -z "$GH_TOC_TOKEN" ]; then
          TOKEN=$GH_TOC_TOKEN
      else
@@ -30,65 +31,221 @@
          if [ -f "$TOKEN_FILE" ]; then
              TOKEN="$(cat $TOKEN_FILE)"
          fi
-@@ -124,12 +122,14 @@ gh_toc(){
+@@ -109,6 +107,15 @@ gh_is_url() {
+     esac
+ }
+ 
++gh_is_invalid_url() {
++    local gh_hostname="github.com"
++
++    case $1 in
++        https://${gh_hostname}/* | http://${gh_hostname}/* ) return 0 ;;
++        *) return 1 ;;
++    esac
++}
++
+ #
+ # TOC generator
+ #
+@@ -139,75 +146,112 @@ gh_toc(){
      fi
  
      if [ "$(gh_is_url "$gh_src")" == "yes" ]; then
--        gh_toc_load "$gh_src" | gh_toc_grab "$gh_src_copy"
+-        gh_toc_load "$gh_src" | gh_toc_grab "$gh_src_copy" "$indent"
 -        if [ "${PIPESTATUS[0]}" != "0" ]; then
-+	local gh_content
-+	gh_content=`gh_toc_load "$gh_src"`
++        if ! which -s "jq"; then
++            echo "Please, install 'jq' and try again."
++            exit 1
++        fi
++
++        if ! gh_is_invalid_url "${gh_src}"; then
++            echo "It looks like an invalid URL."
++            echo "Note that valid URLs are, for example, \"https://github.com/<account>/<repo>\"."
++            exit 1
++        fi
++
++        local account repo
++
++        account=$(echo -n "${gh_src}" | sed -Ee 's#https?://github\.com/([^/]+).+#\1#')
++        repo=$(echo -n "${gh_src}" | sed -Ee 's#https?://github\.com/[^/]+/([^/]+)/?.*#\1#')
++
++        local gh_content
++        gh_content=$(gh_toc_load "${gh_src}")
 +        if [ $? -ne 0 ]; then
              echo "Could not load remote document."
              echo "Please check your url or network connectivity"
              exit 1
          fi
-+        printf "%s\n" "$gh_content" | gh_toc_grab "$gh_src_copy"
++
          if [ "$need_replace" = "yes" ]; then
              echo
              echo "!! '$gh_src' is not a local file"
-@@ -146,7 +146,7 @@ gh_toc(){
-         if [ "$rawhtml" == "XXRateLimitXX" ]; then
-              echo "Parsing local markdown file requires access to github API"
-              echo "Error: You exceeded the hourly limit. See: https://developer.github.com/v3/#rate-limiting"
--             TOKEN_FILE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)/token.txt"
-+             TOKEN_FILE="$(cd "$(dirname "$0")" && pwd)/token.txt"
-              echo "or place GitHub auth token here: ${TOKEN_FILE}"
-              exit 1
+             echo "!! Can't insert the TOC into it."
+             echo
++            exit 1
+         fi
+-    else
+-        local rawhtml=$(gh_toc_md2html "$gh_src" "$skip_header")
+-        if [ "$rawhtml" == "XXNetworkErrorXX" ]; then
+-             echo "Parsing local markdown file requires access to github API"
+-             echo "Please make sure curl is installed and check your network connectivity"
+-             exit 1
++
++        local json_file
++        json_file=$(mktemp -t gh-md-toc)
++        printf "%s\n" "${gh_content}" | grep defaultBranch | sed -Ee 's#<script [^>]+>(.+)</script>#\1#' > "${json_file}"
++
++        local refName path
++
++        refName=$(cat "${json_file}" | jq --raw-output '.props.initialPayload.overview.overviewFiles.[] | select(.preferredFileType == "readme") | .refName' 2> /dev/null)
++        path=$(cat "${json_file}" | jq --raw-output '.props.initialPayload.overview.overviewFiles.[] | select(.preferredFileType == "readme") | .path' 2> /dev/null)
++
++        gh_src=$(mktemp -t gh-md-toc)
++
++        gh_content=`gh_toc_load "https://raw.githubusercontent.com/${account}/${repo}/${refName}/${path}"`
++        if [ $? -ne 0 ]; then
++            echo "Could not load remote document."
++            echo "Please check your url or network connectivity"
++            exit 1
          fi
-@@ -175,11 +175,7 @@ gh_toc(){
-             fi
+-        if [ "$rawhtml" == "XXRateLimitXX" ]; then
+-             echo "Parsing local markdown file requires access to github API"
+-             echo "Error: You exceeded the hourly limit. See: https://developer.github.com/v3/#rate-limiting"
+-             TOKEN_FILE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)/token.txt"
+-             echo "or place GitHub auth token here: ${TOKEN_FILE}"
+-             exit 1
+-        fi
+-        local toc=`echo "$rawhtml" | gh_toc_grab "$gh_src_copy" "$indent"`
+-        echo "$toc"
+-        if [ "$need_replace" = "yes" ]; then
+-            if grep -Fxq "<!--ts-->" "$gh_src" && grep -Fxq "<!--te-->" "$gh_src"; then
+-                echo "Found markers"
+-            else
+-                echo "You don't have <!--ts--> or <!--te--> in your file...exiting"
+-                exit 1
+-            fi
+-            local ts="<\!--ts-->"
+-            local te="<\!--te-->"
+-            local dt=`date +'%F_%H%M%S'`
+-            local ext=".orig.${dt}"
+-            local toc_path="${gh_src}.toc.${dt}"
+-            local toc_createdby="<!-- Created by https://github.com/ekalinin/github-markdown-toc -->"
+-            local toc_footer="<!-- Added by: `whoami`, at: `date` -->"
+-            # http://fahdshariff.blogspot.ru/2012/12/sed-mutli-line-replacement-between-two.html
+-            # clear old TOC
+-            sed -i${ext} "/${ts}/,/${te}/{//!d;}" "$gh_src"
+-            # create toc file
+-            echo "${toc}" > "${toc_path}"
+-            if [ "${no_footer}" != "yes" ]; then
+-                echo -e "\n${toc_createdby}\n${toc_footer}\n" >> "$toc_path"
+-            fi
  
-             # insert toc file
--            if [[ "`uname`" == "Darwin" ]]; then
+-            # insert toc file
+-            if ! sed --version > /dev/null 2>&1; then
 -                sed -i "" "/${ts}/r ${toc_path}" "$gh_src"
 -            else
 -                sed -i "/${ts}/r ${toc_path}" "$gh_src"
 -            fi
-+	    sed -i "" "/${ts}/r ${toc_path}" "$gh_src"
-             echo
-             if [ "${no_backup}" = "yes" ]; then
-                 rm ${toc_path} ${gh_src}${ext}
-@@ -296,14 +292,18 @@ gh_toc_app() {
-     if [ "$1" = '--version' ]; then
-         echo "$gh_toc_version"
-         echo
--        echo "os:     `lsb_release -d | cut -f 2`"
--        echo "kernel: `cat /proc/version`"
--        echo "shell:  `$SHELL --version`"
-+        echo "os:     `uname -rs`"
-+        echo "kernel: `uname -K`"
-+        echo "shell:  $SHELL"
-         echo
--        for tool in curl wget grep awk sed; do
-+        for tool in grep awk; do
-             printf "%-5s: " $tool
-             echo `$tool --version | head -n 1`
-         done
-+	printf "%-5s: " sed
-+	echo "sed (BSD sed)"
-+	printf "%-5s: " fetch
-+	echo "fetch (BSD fetch)"
-         return
+-            echo
+-            if [ "${no_backup}" = "yes" ]; then
+-                rm "$toc_path" "$gh_src$ext"
+-            fi
+-            echo "!! TOC was added into: '$gh_src'"
+-            if [ -z "${no_backup}" ]; then
+-                echo "!! Origin version of the file: '${gh_src}${ext}'"
+-                echo "!! TOC added into a separate file: '${toc_path}'"
++        printf "%s\n" "${gh_content}" > "${gh_src}"
++
++        rm -f "${json_file}"
++    fi
++
++    local rawhtml=$(gh_toc_md2html "$gh_src" "$skip_header")
++    if [ "$rawhtml" == "XXNetworkErrorXX" ]; then
++         echo "Parsing local markdown file requires access to github API"
++         echo "Please make sure curl is installed and check your network connectivity"
++         exit 1
++    fi
++    if [ "$rawhtml" == "XXRateLimitXX" ]; then
++         echo "Parsing local markdown file requires access to github API"
++         echo "Error: You exceeded the hourly limit. See: https://developer.github.com/v3/#rate-limiting"
++         TOKEN_FILE="$(cd "$(dirname "$0}")" && pwd)/token.txt"
++         echo "or place GitHub auth token here: ${TOKEN_FILE}"
++         exit 1
++    fi
++    local toc=`echo "$rawhtml" | gh_toc_grab "$gh_src_copy" "$indent"`
++    echo "$toc"
++    if [ "$need_replace" = "yes" ]; then
++        if grep -Fxq "<!--ts-->" "$gh_src" && grep -Fxq "<!--te-->" "$gh_src"; then
++            echo "Found markers"
++        else
++            echo "You don't have <!--ts--> or <!--te--> in your file...exiting"
++            exit 1
+         fi
+-            echo
++        local ts="<\!--ts-->"
++        local te="<\!--te-->"
++        local dt=`date +'%F_%H%M%S'`
++        local ext=".orig.${dt}"
++        local toc_path="${gh_src}.toc.${dt}"
++        local toc_createdby="<!-- Created by https://github.com/ekalinin/github-markdown-toc -->"
++        local toc_footer="<!-- Added by: `whoami`, at: `date` -->"
++        # http://fahdshariff.blogspot.ru/2012/12/sed-mutli-line-replacement-between-two.html
++        # clear old TOC
++        sed -i${ext} "/${ts}/,/${te}/{//!d;}" "$gh_src"
++        # create toc file
++        echo "${toc}" > "${toc_path}"
++        if [ "${no_footer}" != "yes" ]; then
++            echo -e "\n${toc_createdby}\n${toc_footer}\n" >> "$toc_path"
+         fi
++
++        # insert toc file
++        sed -i "" "/${ts}/r ${toc_path}" "$gh_src"
++        echo
++        if [ "${no_backup}" = "yes" ]; then
++            rm "$toc_path" "$gh_src$ext"
++        fi
++        echo "!! TOC was added into: '$gh_src'"
++        if [ -z "${no_backup}" ]; then
++            echo "!! Origin version of the file: '${gh_src}${ext}'"
++            echo "!! TOC added into a separate file: '${toc_path}'"
      fi
++        echo
++    fi
+ }
+ 
+ #
+@@ -218,7 +262,6 @@ gh_toc_grab() {
+ # $2 - number of spaces used to indent.
+ #
+ gh_toc_grab() {
+-
+     href_regex="/href=\"[^\"]+?\"/"
+     common_awk_script='
+                      modified_href = ""
+@@ -298,19 +341,21 @@ show_version() {
+ show_version() {
+     echo "$gh_toc_version"
+     echo
+-    echo "os:     `uname -s`"
++    echo "os:     `uname -rs`"
+     echo "arch:   `uname -m`"
+-    echo "kernel: `uname -r`"
+-    echo "shell:  `$SHELL --version`"
++    echo "kernel: `uname -i`"
++    echo "shell:  $SHELL"
+     echo
+-    for tool in curl wget grep awk sed; do
++    for tool in curl grep awk jq; do
+         printf "%-5s: " $tool
+-        if `type $tool &>/dev/null`; then
++        if type $tool > /dev/null 2>&1; then
+             echo `$tool --version | head -n 1`
+         else
+             echo "not installed"
+         fi
+     done
++    printf "%-5s: " sed
++    echo "sed (BSD sed)"
+ }
  
+ show_help() {