Add awk script for gemtext to man conversion

2 files changed, 127 insertions, 95 deletions

diff --git a/doc/gem-to-man.awk b/doc/gem-to-man.awk
new file mode 100755
index 0000000..dd1f590
--- /dev/null
+++ b/doc/gem-to-man.awk
@@ -0,0 +1,83 @@
+#!/bin/awk -f
+
+BEGIN {
+    preformatted=0
+}
+
+# Empty lines
+/^$/ {
+    print "."
+    next
+}
+
+# Level 2 headings become section headings.
+/^##\s/ {
+    print ".SH"
+    sub(/^##\s/, "")
+    print toupper($0)
+    next
+}
+
+# Level 3 headings become subsection headings.
+/^###\s/ {
+    sub(/^###\s/, "")
+    print ".SS \"" $0 "\""
+    next
+}
+
+# Lists
+/^\*\s/ {
+    print ".IP \\(bu 3"
+    sub(/^\*\s/, "")
+    print
+    next
+}
+
+# Preformatted text
+/^```/ {
+    # We simply place indent macros to make
+    # it stand out a little.
+    if (!preformatted)
+    {
+        print ".RS"
+        preformatted=1
+    }
+    else
+    {
+        preformatted=0
+        print ".RE"
+    }
+    next
+}
+
+# Links
+/^=>/ {
+    # Strips => prefix, and separates URL and link
+    # title.
+    sub(/^=>/, "")
+    url=$1
+    $1=""
+    title=$0
+
+    print ".IP"
+
+    # Print either URL on its own or Title: URL if we
+    # have a title.
+    if (title == "")
+    {
+        print url
+    }
+    else
+    {
+        gsub(/^\s/, "", title)
+        print title ": " url
+    }
+
+    next
+}
+
+# Paragraphs (default)
+{
+    print ".PP"
+    print
+}
diff --git a/doc/gen-man.sh b/doc/gen-man.sh
index 6a48cc8..c8e6fdf 100755
--- a/doc/gen-man.sh
+++ b/doc/gen-man.sh
@@ -17,106 +17,55 @@ man_head="./kristall-head.man"
 man_tail="./kristall-tail.man"
 man_output="./kristall.1"
 gemtext_in="../src/about/help.gemini"
+gemtext_converter="./gem-to-man.awk"
 
-main() {
-    # Make sure we have all the stuff we need
-    if [[ ! -f "$man_head" ]]; then
-        echo "man page head does not exist."
-        exit -1
-    fi
-    if [[ ! -f "$man_tail" ]]; then
-        echo "man page tail does not exist"
-        exit -1
-    fi
-    if [[ -z "$man_output" ]]; then
-        echo "No output file"
-        exit -1
-    fi
-    if [[ ! -f "$gemtext_in" ]]; then
-        echo "Input gemtext file does not exist"
-        exit -1
-    fi
+# Make sure we have all the stuff we need
+if [[ ! -f "$man_head" ]]; then
+    echo "man page head does not exist."
+    exit -1
+fi
+if [[ ! -f "$man_tail" ]]; then
+    echo "man page tail does not exist"
+    exit -1
+fi
+if [[ -z "$man_output" ]]; then
+    echo "No output file"
+    exit -1
+fi
+if [[ ! -f "$gemtext_in" ]]; then
+    echo "Input gemtext file does not exist"
+    exit -1
+fi
+if [[ ! -f "$gemtext_converter" ]]; then
+    echo "Gemtext converter script does not exist"
+    exit -1
+fi
 
-    # Write the head to the output file.
-    cp "$man_head" "$man_output"
+# Write the head to the output file.
+cp "$man_head" "$man_output"
 
-    # Insert last modified date (use last-modified date of help.gemini)
-    last_modified=$(date -r "$gemtext_in" +"%Y-%M-%d")
-    sed -i "$man_output" -e 's#\$(DATE)#'"$last_modified"'#g'
+# Insert last modified date (use last-modified date of help.gemini)
+last_modified=$(date -r "$gemtext_in" +"%Y-%M-%d")
+sed -i "$man_output" -e 's#\$(DATE)#'"$last_modified"'#g'
 
-    # Convert gemtext about page to roff's man format:
-    convert_gemtext "$gemtext_in" >> "$man_output"
+# Some pre-processing before giving our gemtext to the awk script.
+gem_in=$(
+    # Read input file
+    cat "$gemtext_in" |
 
-    # Write the tail to the output file.
-    cat "$man_tail" >> "$man_output"
-}
+    # Strip a few lines from beginning/end of file.
+    tail -n +7 |
+    head -n -7 |
 
-# Simple converter for gemtext to roff
-# Note that this is not a converter of "standard" gemtext, and is
-# designed specifically for kristall's about page. Thus, this may not work
-# as expected on other gemtext files (may require a little tweaking)
-convert_gemtext() {
-    gem_in="$(cat $1 | tail -n +7 | head -n -7)"
+    # First expression replaces all [Text like this] with bold text.
+    # Second expression replaces text like *This* or _this_ with italic text.
+    sed -E \
+      -e 's#\[([^]]*)\]#\\fB\1\\fR#g' \
+      -e 's#(^|[.,!? ]+)[*_]([^*_ ]+[^*_]+[^*_ ]+)[*_]($|[.,!? ])#\1\\fI\2\\fR\3#g'
+)
 
-    # Replace [Text] with bold text.
-    gem_in="$(echo "$gem_in" | sed -Ee 's#\[([^]]*)\]#\\fB\1\\fR#g')"
+# Convert gemtext to man format
+echo "$gem_in" | "$gemtext_converter" >> "$man_output"
 
-    # Replace *Text* or _text_ with italic (not bold to help differentiate with the above)
-    # (Regex is derived from one in geminirenderer.cpp)
-    gem_in="$(echo "$gem_in" | sed -Ee 's#(^|[.,!? ]+)[*_]([^*_ ]+[^*_]+[^*_ ]+)[*_]($|[.,!? ])#\1\\fI\2\\fR\3#g')"
-
-    is_preformatted=false
-
-    # Iterate over all the lines in the gemtext file
-    while IFS= read -r line; do
-        if echo "$line" | grep -q '^##\s'; then
-            # Top level heading
-            text="${line#'## '}"
-            echo ".SH"
-            echo "${text^^}"
-        elif echo "$line" | grep -q '^###\s'; then
-            # Level 2 heading
-            text="${line#'### '}"
-            echo ".SS"
-            echo "$text"
-        elif echo "$line" | grep -q '^\*\s'; then
-            # Lists
-            text="${line#'* '}"
-            echo ".IP \(bu 3"
-            echo "${text}"
-        elif echo "$line" | grep -q '^\=>'; then
-            # Links. We only show the URL itself if there is one.
-            text="${line#'=>'}"
-            title="$(echo "$text" | awk '{$1=""; print $0}')"
-            text="$(echo "$text" | awk '{print $1}')"
-
-            # Prefix with URL title if we have one
-            if [[ -n "$title" ]]; then
-                text="$title: $text"
-            fi
-
-            echo ".IP"
-
-            # Echo the text, we also strip leading whitespace
-            echo "$text" | sed -e 's#^[[:space:]]*##'
-        elif echo "$line" | grep -q '^```'; then
-            # Preformatted text. We just indent it
-            if [[ "$is_preformatted" = false ]]; then
-                echo ".RS"
-                is_preformatted=true
-            else
-                echo ".RE"
-                is_preformatted=false
-            fi
-        elif [[ -z "$line" ]]; then
-            # Empty lines
-            echo "."
-        else
-            # Regular paragraphs
-            echo ".PP"
-            echo "$line"
-        fi
-    done <<< "$gem_in"
-}
-
-main "$@"
+# Write the tail to the output file.
+cat "$man_tail" >> "$man_output"