mirror of
https://github.com/RubyMetric/chsrc
synced 2025-07-18 05:27:38 +08:00
Basic implementation of rawstr4c
This commit is contained in:
Aoran Zeng
parent
747c29e85f
commit
a54f704719
@ -7,6 +7,10 @@
|
|||||||
# Last Modified : <2025-07-12>
|
# Last Modified : <2025-07-12>
|
||||||
#
|
#
|
||||||
# Generate raw strings for C programming language
|
# Generate raw strings for C programming language
|
||||||
|
#
|
||||||
|
# ---------------------------------------------------------------
|
||||||
|
# 全局设置称为 Global config，其他均称为 section config
|
||||||
|
# 每一部分称为 section，我们要处理的input，都在 code block 里，我们简称为 block
|
||||||
# ---------------------------------------------------------------
|
# ---------------------------------------------------------------
|
||||||
|
|
||||||
unless @*ARGS {
|
unless @*ARGS {
|
||||||
@ -15,37 +19,20 @@ unless @*ARGS {
|
|||||||
|
|
||||||
my $markdown-file = @*ARGS[0];
|
my $markdown-file = @*ARGS[0];
|
||||||
|
|
||||||
# 确保文件存在
|
|
||||||
unless $markdown-file.IO.e {
|
unless $markdown-file.IO.e {
|
||||||
die "Error: File '$markdown-file' not found.\n";
|
die "Error: File '$markdown-file' not found.\n";
|
||||||
}
|
}
|
||||||
|
|
||||||
# 解析配置信息
|
|
||||||
sub parse-config($content) {
|
|
||||||
my %config;
|
|
||||||
|
|
||||||
for $content.lines -> $line {
|
|
||||||
# say "Processing line: $line";
|
|
||||||
if $line ~~ /^ '-' \s* 'translate' \s* '=' \s* '`' ':' (<[a..z\-]>+) '`' / {
|
|
||||||
%config<translation> = ~$0;
|
|
||||||
say "Global translation mode: " ~ %config<translation>;
|
|
||||||
last;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
return %config;
|
|
||||||
}
|
|
||||||
|
|
||||||
# 根据转换模式处理字符
|
# 根据转换模式处理字符
|
||||||
sub convert-char($char, $mode) {
|
sub convert-char($char, $mode) {
|
||||||
my $byte = $char.encode('UTF-8')[0];
|
|
||||||
|
|
||||||
given $mode {
|
given $mode {
|
||||||
when 'octal' {
|
when 'oct' {
|
||||||
return "\\" ~ sprintf("%03o", $byte);
|
my $bytes = $char.encode('UTF-8');
|
||||||
|
return $bytes.map({ "\\" ~ sprintf("%03o", $_) }).join('');
|
||||||
}
|
}
|
||||||
when 'hex' {
|
when 'hex' {
|
||||||
return "\\x" ~ sprintf("%02x", $byte);
|
my $bytes = $char.encode('UTF-8');
|
||||||
|
return $bytes.map({ "\\x" ~ sprintf("%02x", $_) }).join('');
|
||||||
}
|
}
|
||||||
when 'escape' {
|
when 'escape' {
|
||||||
# 只转义必要的字符
|
# 只转义必要的字符
|
||||||
@ -66,24 +53,170 @@ sub convert-char($char, $mode) {
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
# 处理字符串转换
|
||||||
|
sub process-content($content, $mode) {
|
||||||
|
my $result = "";
|
||||||
|
for $content.comb -> $char {
|
||||||
|
$result ~= convert-char($char, $mode);
|
||||||
|
}
|
||||||
|
return $result;
|
||||||
|
}
|
||||||
|
|
||||||
|
# 生成变量名
|
||||||
|
sub generate-variable-name($global-config, $section-config, $title) {
|
||||||
|
my $prefix = $global-config<prefix> // "_rawstr4c";
|
||||||
|
my $postfix = $global-config<postfix> // "";
|
||||||
|
|
||||||
|
# 处理前缀
|
||||||
|
$prefix = $prefix.subst(/^'`'/, '').subst(/'`'$/, '');
|
||||||
|
|
||||||
|
# 处理后缀
|
||||||
|
if $postfix {
|
||||||
|
$postfix = $postfix.subst(/^':'/, '');
|
||||||
|
if $section-config<language> {
|
||||||
|
my $lang = $section-config<language>.subst(/^'`'/, '').subst(/'`'$/, '');
|
||||||
|
$postfix = $postfix.subst('use-language', "in_$lang");
|
||||||
|
} else {
|
||||||
|
$postfix = $postfix.subst('use-language', ''); # 默认为无语言
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
# 生成的变量名称
|
||||||
|
my $name = $section-config<name> // $title.lc;
|
||||||
|
$name = $name.subst(/^'`'/, '').subst(/'`'$/, '');
|
||||||
|
# 处理标题中包含的空格
|
||||||
|
$name = $name.subst(/\s+/, '_', :g);
|
||||||
|
|
||||||
|
my $var-name = $prefix;
|
||||||
|
if $name {
|
||||||
|
$var-name ~= "_" ~ $name;
|
||||||
|
}
|
||||||
|
|
||||||
|
if $postfix {
|
||||||
|
$var-name ~= "_" ~ $postfix;
|
||||||
|
}
|
||||||
|
|
||||||
|
return $var-name;
|
||||||
|
}
|
||||||
|
|
||||||
|
#`( 真正的 main 流程开始
|
||||||
|
)
|
||||||
my $content = $markdown-file.IO.slurp;
|
my $content = $markdown-file.IO.slurp;
|
||||||
|
my @lines = $content.lines;
|
||||||
|
|
||||||
my %config = parse-config($content);
|
my %global-config;
|
||||||
my $translation-mode = %config<translation> // 'octal';
|
my @sections;
|
||||||
|
my $current-section;
|
||||||
|
my $in-global = True;
|
||||||
my $in-code-block = False;
|
my $in-code-block = False;
|
||||||
for $content.lines -> $line {
|
my $current-code = "";
|
||||||
if $line ~~ /^ '```' / {
|
|
||||||
# 遇到代码块的开始或结束标记
|
for @lines -> $line {
|
||||||
$in-code-block = !$in-code-block;
|
# 检查标题级别 (注意，要确保不在代码块内)
|
||||||
next; # 跳过代码块标记行
|
if !$in-code-block && $line ~~ /^ '#' ('#'*) \s* (.+) / {
|
||||||
|
my $level = 1 + $0.chars;
|
||||||
|
my $title = ~$1;
|
||||||
|
|
||||||
|
# 如果之前有代码块，保存
|
||||||
|
if $current-code && $current-section && $current-section<title> {
|
||||||
|
$current-section<code> = $current-code;
|
||||||
|
@sections.push: $current-section;
|
||||||
|
}
|
||||||
|
|
||||||
|
# 重置代码块
|
||||||
|
$current-code = "";
|
||||||
|
|
||||||
|
# 一级标题后面是全局配置，其他级别标题开始新的section
|
||||||
|
if $level == 1 {
|
||||||
|
$in-global = True;
|
||||||
|
$current-section = {};
|
||||||
|
} else {
|
||||||
|
$in-global = False;
|
||||||
|
$current-section = {
|
||||||
|
title => $title,
|
||||||
|
level => $level,
|
||||||
|
config => {},
|
||||||
|
};
|
||||||
|
}
|
||||||
|
next;
|
||||||
|
}
|
||||||
|
|
||||||
|
# 解析配置项
|
||||||
|
if $line ~~ /^ '-' \s* (<[a..z\-]>+) \s* '=' \s* '`' (.+?) '`' / {
|
||||||
|
my $key = ~$0;
|
||||||
|
my $value = ~$1;
|
||||||
|
|
||||||
|
if $in-global {
|
||||||
|
%global-config{$key} = $value;
|
||||||
|
} elsif $current-section {
|
||||||
|
$current-section<config>{$key} = $value;
|
||||||
|
}
|
||||||
|
next;
|
||||||
|
}
|
||||||
|
|
||||||
|
# 处理代码块，即真正想要转换的 raw string
|
||||||
|
# 检测代码块语言并记录
|
||||||
|
if $line ~~ /^ '```' (.*)? / {
|
||||||
|
if $in-code-block {
|
||||||
|
# 代码块结束
|
||||||
|
$in-code-block = False;
|
||||||
|
} else {
|
||||||
|
# 代码块开始，记录语言
|
||||||
|
$in-code-block = True;
|
||||||
|
my $lang = ~($0 // '');
|
||||||
|
if $lang && $current-section {
|
||||||
|
# 如果代码块指定了语言，且当前section没有language配置，则自动设置
|
||||||
|
unless $current-section<config><language> {
|
||||||
|
$current-section<config><language> = $lang;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
next;
|
||||||
}
|
}
|
||||||
|
|
||||||
if $in-code-block {
|
if $in-code-block {
|
||||||
# 如果在代码块内部，处理每一行
|
$current-code ~= $line ~ "\n";
|
||||||
for $line.comb -> $char {
|
|
||||||
print convert-char($char, $translation-mode);
|
|
||||||
}
|
|
||||||
print convert-char("\n", $translation-mode); # 处理换行符
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
# 保存最后一个代码块
|
||||||
|
if $current-code && $current-section && $current-section<title> {
|
||||||
|
$current-section<code> = $current-code;
|
||||||
|
@sections.push: $current-section;
|
||||||
|
}
|
||||||
|
|
||||||
|
# 解析完毕，最后输出结果
|
||||||
|
say "Global config:";
|
||||||
|
for %global-config.kv -> $k, $v {
|
||||||
|
say " $k = $v";
|
||||||
|
}
|
||||||
|
say "";
|
||||||
|
|
||||||
|
say "Found " ~ @sections.elems ~ " sections:";
|
||||||
|
for @sections -> $section {
|
||||||
|
say "Section: " ~ $section<title>;
|
||||||
|
}
|
||||||
|
say "";
|
||||||
|
|
||||||
|
for @sections -> $section {
|
||||||
|
say "=== Section: " ~ $section<title> ~ " ===";
|
||||||
|
|
||||||
|
# 确定转换模式 (section config 优先，否则使用 global config)
|
||||||
|
my $translate = $section<config><translate> // %global-config<translate> // ':oct';
|
||||||
|
$translate = $translate.subst(/^':'/, '');
|
||||||
|
|
||||||
|
# 生成变量名
|
||||||
|
my $var-name = generate-variable-name(%global-config, $section<config>, $section<title>);
|
||||||
|
|
||||||
|
say "Variable name: $var-name";
|
||||||
|
say "Translation mode: $translate";
|
||||||
|
|
||||||
|
my $language = $section<config><language>;
|
||||||
|
say "Language: $language";
|
||||||
|
|
||||||
|
if $section<code> {
|
||||||
|
say '';
|
||||||
|
say 'char ' ~ $var-name ~ '[] = "' ~ process-content($section<code>, $translate) ~ '";';
|
||||||
|
}
|
||||||
|
say "\n";
|
||||||
|
}
|
||||||
|
|||||||
Loading…
x
Reference in New Issue
Block a user