Improve multi-language support

2022-07-02 20:55:45 +00:00 · 2022-07-02 20:55:45 +00:00 · 6c831dc78f
commit 6c831dc78f
parent 121ebca9ae
3 changed files with 27 additions and 9 deletions
--- a/README.md
+++ b/README.md
@ -7,7 +7,6 @@ Many people don't like to read markdown.

 ## Usage

-These are AWK scripts.
 The input markdown goes into stdin.
 The output document comes out of stdout.

@ -24,12 +23,12 @@ pdflatex README.tex
 If the awk option `stylesheet` is given, the HTML document can made prettier with external CSS.

 ```
-awk -v stylesheet='/style.css' -f md2html < README.md
+md2html -v stylesheet='/style.css' < README.md
 ```

 ## Supported constructs

- Headlines up to level 3
+- Headlines up to level 6
 - Paragraphs
 - Bold, italic and inline monospace
 - Image figures if they are in the same folder
@ -40,7 +39,6 @@ awk -v stylesheet='/style.css' -f md2html < README.md
 ## Not supported

 - Tables
- More than 3 levels of nested headlines
 - Images from an url
- Inline images
- Unicode characters (im german, maybe i will add it for umlauts)
+- Images inline in text
+- Unicode characters (im german, maybe i will add at least umlaut support)
--- a/16
+++ b/16
@ -7,6 +7,7 @@ BEGIN {
  prefix = "html/body/"
  print("<!DOCTYPE html>")
  level("html/head")
+  meta["lang"]="en"
  toc=0
 }

@ -46,7 +47,11 @@ function level(new) {

  for (i = common; i < length(a); i++) {
    if (a[i]) {
+      if (a[i]=="body") {
+        printf("%0" indent "s<%s lang=%s>\n", "", a[i], meta["lang"])
+      } else {
        printf("%0" indent "s<%s>\n", "", a[i])
+      }
      indent = indent + 2
    }
  }
@ -188,6 +193,9 @@ awk '
 BEGIN {
  indent = 4
  level("summary")
+  lang="en"
+  tocname["en"]="Table of contents"
+  tocname["de"]="Inhaltsverzeichnis"
 }

 function oneliner(tag, str, args) {
@ -242,6 +250,14 @@ function inline(text) {
  return text
 }

+/<body lang=/ {
+  lang=$0
+  gsub(/^.*lang=\"*/,"",lang)
+  gsub(/[^a-z].*$/,"",lang)
+  level("summary")
+  oneliner("h2", tocname[lang])
+}
+
 /<h[2-6] / {
  id=$0
  gsub(/^.*id=\"/,"",id)
--- a/8
+++ b/8
@ -5,6 +5,7 @@ BEGIN {
  tag("documentclass[a4paper]{article}")
 #  tag("special{papersize=210mm,297mm}")
  tag("usepackage[margin=1in]{geometry}")
+  tag("usepackage[utf8]{inputenc}")
  tag("setlength{\\parskip}{\\baselineskip}%")
  tag("setlength{\\parindent}{0pt}%")
  tag("usepackage{listings}")
@ -82,9 +83,12 @@ function inline(text) {
  }
 }

-/^## / {
+/^##* / {
+  d=length($1)
+  t="section"
+  for (i=3; i<=d; i++) t="sub"t
  gsub(/^#* /, "", $0)
-  tag("section{" $0 "}")
+  tag(t"{" $0 "}")
  next
 }