view mrjunejune/create_html_from_md.c @ 179:8d17f6e6e290

[ThirdParty] Added emsdk bazel rules that can be supported by bazel 9.0.0
author MrJuneJune <me@mrjunejune.com>
date Thu, 22 Jan 2026 21:23:17 -0800
parents 295ac2e5ec00
children
line wrap: on
line source

#include "markdown_converter/markdown_to_html.h"
#include "dowa/dowa.h"
#include <fcntl.h>

#define BLOG_PATH "/home/june/zenbu/mrjunejune/src/blog/"
#define BLOG_HTML "<!DOCTYPE html>" \
"<html lang=\"en\">" \
"<head>" \
"    <meta charset=\"UTF-8\">" \
"    <meta name=\"viewport\" content=\"width=device-width, initial-scale=1.0\">" \
"    <meta name=\"description\" content=\"%s\">" \
"    <title>%s - June's Blog</title>" \
"    {{/parts/base_head.html}}" \
"    <link rel=\"stylesheet\" href=\"/public/highlight/a11y-dark.min.css\"  media=\"(prefers-color-scheme: dark)\">" \
"    <link rel=\"stylesheet\" href=\"/public/highlight/a11y-light.min.css\" media=\"(prefers-color-scheme: light)\">" \
"    <script src=\"/public/highlight/highlight.min.js\"></script>" \
"</head>" \
"<body>" \
"  {{/parts/header.html}}" \
"  <main>" \
"%s" \
"  </main>" \
"  {{/parts/footer.html}}" \
"  <script>hljs.highlightAll();</script>"\
"</body>" 


typedef struct {
  char title[256];
  char description[512];
  char *content;       // markdown content after frontmatter
} Blog_Metadata;

// Parse frontmatter from markdown content
// Expected format:
// ---
// title: My Title
// description: My description
// ---
// markdown
static Blog_Metadata Parse_Blog_Frontmatter(const char *md_content)
{
  Blog_Metadata meta = {0};
  strcpy(meta.title, "Untitled");
  strcpy(meta.description, "A blog post by June");
  meta.content = (char *)md_content;

  if (!md_content) return meta;

  if (strncmp(md_content, "---", 3) != 0)
    return meta;

  const char *end_delim = strstr(md_content + 3, "\n---");
  if (!end_delim)
    return meta;

  const char *cursor = md_content + 4; // skip "---\n"
  while (cursor < end_delim)
  {
    // whitespaces
    while (cursor < end_delim && (*cursor == ' ' || *cursor == '\t')) cursor++;

    const char *line_end = cursor;
    while (line_end < end_delim && *line_end != '\n') line_end++;

    const char *colon = cursor;
    while (colon < line_end && *colon != ':') colon++;

    if (colon < line_end) {
      size_t key_len = colon - cursor;
      const char *value_start = colon + 1;
      while (value_start < line_end && (*value_start == ' ' || *value_start == '\t')) value_start++;
      size_t value_len = line_end - value_start;

      if (key_len == 5 && strncmp(cursor, "title", 5) == 0)
      {
        size_t copy_len = value_len < sizeof(meta.title) - 1 ? value_len : sizeof(meta.title) - 1;
        strncpy(meta.title, value_start, copy_len);
        meta.title[copy_len] = '\0';
      }
      else if (key_len == 11 && strncmp(cursor, "description", 11) == 0)
      {
        size_t copy_len = value_len < sizeof(meta.description) - 1 ? value_len : sizeof(meta.description) - 1;
        strncpy(meta.description, value_start, copy_len);
        meta.description[copy_len] = '\0';
      }
    }

    cursor = line_end + 1;
  }

  meta.content = (char *)(end_delim + 4); 
  if (*meta.content == '\n') meta.content++;

  return meta;
}

int main()
{
  Dowa_Arena *arena = Dowa_Arena_Create(1024 * 1024 * 5);
  size_t buffer_sizes = 100 * 1024;

  char *files[] = {
    BLOG_PATH"thoughts-on-ide/index.md",
    BLOG_PATH"multithread-in-js/index.md",
    BLOG_PATH"thoughts-on-tdd/index.md",
    BLOG_PATH"my-seobeo-journey/index.md",
    BLOG_PATH"wasm-bunny/index.md",
    BLOG_PATH"optimizing-data-structures/index.md",
    BLOG_PATH"websocket-demystified/index.md",
    BLOG_PATH"optimizing-grass-rendering/index.md",
    BLOG_PATH"wsl2-ssh/index.md"
  };

  for (int i = 0; i < 9; i++)
  {
    char *md_file_path = files[i];
    char *md_file = Dowa_Arena_Allocate(arena, buffer_sizes);

    FILE *file = fopen(md_file_path, "ro");
    fseek(file, 0, SEEK_END);
    size_t file_length = ftell(file);
    fseek(file, 0, SEEK_SET);
    fread(md_file, 1, file_length, file);

    Blog_Metadata meta = Parse_Blog_Frontmatter(md_file);
    char *md = markdown_to_html(meta.content);
    char *ssr_body = Dowa_Arena_Allocate(arena, buffer_sizes);
    char *final_body = Dowa_Arena_Allocate(arena, buffer_sizes);
    snprintf(ssr_body, buffer_sizes, BLOG_HTML, meta.description, meta.title, md);

    int open_flags = O_RDWR | O_CREAT | O_EXCL;
    size_t md_file_path_length = strlen(md_file_path);
    md_file_path_length -= 3; // html
    char *new_file_path = Dowa_Arena_Allocate(arena, 1024);
    snprintf(new_file_path, 1024, "%.*s.html", (int)md_file_path_length, md_file_path);

    FILE *new_file_fd = fopen(new_file_path, "w+");
    if (!new_file_fd)
      return;
    printf("Saving to %s...\n", new_file_path);
    fwrite(ssr_body, 1, buffer_sizes, new_file_fd);
    fclose(new_file_fd);
  }
  return 0;
}