kc3-lang/kc3/pdf/pdf_buf_parse.c

• Show log

  Commit

• Hash : 79484957
  Author : lyzer
  Date : 2025-09-15T16:33:38
  

  feat: pdf header parser
  

Download

• pdf/pdf_buf_parse.c

• /* kc3
   * Copyright from 2022 to 2025 kmx.io <contact@kmx.io>
   *
   * Permission is hereby granted to use this software granted the above
   * copyright notice and this permission paragraph are included in all
   * copies and substantial portions of this software.
   *
   * THIS SOFTWARE IS PROVIDED "AS-IS" WITHOUT ANY GUARANTEE OF
   * PURPOSE AND PERFORMANCE. IN NO EVENT WHATSOEVER SHALL THE
   * AUTHOR BE CONSIDERED LIABLE FOR THE USE AND PERFORMANCE OF
   * THIS SOFTWARE.
   */
  #include "../libkc3/kc3.h"
  #include "pdf_buf_parse.h"
  #include "pdf_file.h"
  
  // TODO: use xref instead
  sw pdf_buf_ignore_until_token (s_buf *buf, const char *token)
  {
    sw r;
    sw result = 0;
    assert(buf);
    assert(token);
    while (1) {
      if ((r = pdf_buf_parse_token(buf, token)) < 0)
        return r;
      if (r) {
        result += r;
        break;
      }
      if ((r = buf_ignore(buf, 1)) <= 0)
        return r;
      result += r;
    }
    return result;
  }
  
  sw pdf_buf_parse (s_buf *buf, s_tag *dest)
  {
    sw r;
    sw result = 0;
    s_tag tmp = {0};
    if ((r = buf_ignore_spaces(buf)) < 0)
      return r;
    result += r;
    if ((r = buf_parse_comments(buf)) < 0)
      goto ok;
    result += r;
    if ((r = pdf_buf_parse_indirect_object(buf, &tmp.data.tuple)) > 0) {
      result += r;
      tmp.type = TAG_TUPLE;
      goto ok;
    }
    if ((r = pdf_buf_parse_bool(buf, &tmp.data.bool_)) > 0) {
      result += r;
      tmp.type = TAG_BOOL;
      goto ok;
    }
    if ((r = pdf_buf_parse_null(buf, &tmp)) > 0) {
      result += r;
      goto ok;
    }
    if ((r = pdf_buf_parse_number(buf, &tmp)) > 0) {
      result += r;
      goto ok;
    }
    if ((r = pdf_buf_parse_dictionnary(buf, &tmp.data.map)) > 0) {
      result += r;
      tmp.type = TAG_MAP;
      goto ok;
    }
    if ((r = pdf_buf_parse_string(buf, &tmp)) > 0) {
      result += r;
      goto ok;
    }
    if ((r = pdf_buf_parse_name(buf, &tmp.data.psym)) > 0) {
      tmp.type = TAG_PSYM;
      result += r;
      goto ok;
    }
    if ((r = pdf_buf_parse_array(buf, &tmp.data.plist)) > 0) {
      tmp.type = TAG_PLIST;
      result += r;
      goto ok;
    }
   ok:
    if (result)
      *dest = tmp;
    return result;
  }
  
  sw pdf_buf_parse_array (s_buf *buf, p_list *dest)
  {
    bool end;
    sw r;
    sw result = 0;
    p_list *tail;
    p_list tmp = NULL;
    s_buf_save save = {0};
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, "[")) <= 0)
      goto clean;
    result += r;
    tail = &tmp;
    while (1) {
      if ((r = buf_read_1(buf, "]")) < 0)
        goto restore;
      result += r;
      if (r) {
        result += r;
        if ((r = pdf_buf_parse_object_end(buf, &end)) > 0)
          result += r;
        if (! end)
          goto restore;
        break;
      }
      if (! (*tail = list_new(NULL))) {
        r = -2;
        goto restore;
      }
      if ((r = pdf_buf_parse(buf, &(*tail)->tag)) <= 0)
        goto restore;
      result += r;
      tail = &(*tail)->next.data.plist;
    }
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    list_delete_all(tmp);
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_bool (s_buf *buf, bool *dest)
  {
    bool end;
    sw r;
    sw result = 0;
    s_buf_save save = {0};
    bool tmp;
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, "true")) > 0)
      tmp = true;
    else if ((r = buf_read_1(buf, "false")) > 0)
      tmp = false;
    else
      goto clean;
    result += r;
    if ((r = pdf_buf_parse_object_end(buf, &end)) > 0)
      result += r;
    if (! end) {
      r = 0;
      goto restore;
    }
    result += r;
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_comment (s_buf *buf)
  {
    sw r;
    sw result = 0;
    s_str str = {0};
    assert(buf);
    if ((r = buf_read_1(buf, "%")) <= 0)
      return r;
    result += r;
    if ((r = buf_read_until_1_into_str(buf, "\n", &str)) < 0)
      return result;
    str_clean(&str);
    return result;
  }
  
  sw pdf_buf_parse_comments (s_buf *buf)
  {
    sw r1 = 1;
    sw r2 = 1;
    sw result = 0;
    assert(buf);
    while (r1 | r2) {
      if ((r1 = pdf_buf_parse_comment(buf)) < 0)
        return r1;
      result += r1;
      if ((r2 = buf_ignore_spaces(buf)) < 0)
        return r2;
      result += r2;
    }
    return result;
  }
  
  sw pdf_buf_parse_dictionnary (s_buf *buf, s_map *dest)
  {
    p_list  keys = NULL;
    p_list *keys_tail = &keys;
    const s_sym *name;
    sw r;
    sw result = 0;
    s_buf_save save;
    s_map tmp = {0};
    p_list  values = NULL;
    p_list *values_tail = &values;
    buf_save_init(buf, &save);
    if ((r = pdf_buf_parse_token(buf, "<<")) <= 0)
      goto clean;
    result += r;
    while (1) {
      if ((r = pdf_buf_parse_token(buf, ">>")) < 0)
        goto clean;
      if (r) {
        result += r;
        goto ok;
      }
      if ((r = pdf_buf_parse_name(buf, &name)) <= 0)
        goto clean;
      result += r;
      if (! (*keys_tail = list_new_psym(name, NULL))) {
        r = -1;
        goto clean;
      }
      if (! (*values_tail = list_new(NULL))) {
        r = -1;
        goto clean;
      }
      if ((r = pdf_buf_parse(buf, &(*values_tail)->tag)) <= 0)
        goto clean;
      result += r;
      keys_tail = &(*keys_tail)->next.data.plist;
      values_tail = &(*values_tail)->next.data.plist;
    }
   ok:
    if (! map_init_from_lists(&tmp, keys, values)) {
      err_puts("pdf_buf_parse_dictionnary: map_init_from_lists");
      assert(! "pdf_buf_parse_dictionnary: map_init_from_lists");
      r = -1;
      goto restore;
    }
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    list_delete_all(values);
    list_delete_all(keys);
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_file_header (s_buf *buf, s_str *dest)
  {
    u16 i;
    sw r;
    sw result = 0;
    s_buf_save save;
    s_str tmp = {0};
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, "%PDF-")) <= 0) 
      goto clean;
    result += r;
    if ((r = buf_parse_u16_decimal(buf, &i)) <= 0 || i < 1) {
      err_puts("pdf_buf_parse_file_header: buf_parse_u16_decimal 1");
      assert(! "pdf_buf_parse_file_header: buf_parse_u16_decimal 1");
      goto restore;
    }
    result += r;
    if ((r = buf_read_1(buf, ".")) <= 0) {
      err_puts("pdf_buf_parse_file_header: buf_read_1 2");
      assert(! "pdf_buf_parse_file_header: buf_read_1 2");
      goto restore;
    }
    result += r;
    if ((r = buf_parse_u16_decimal(buf, &i)) <= 0) {
      err_puts("pdf_buf_parse_file_header: buf_parse_u16_decimal");
      assert(! "pdf_buf_parse_file_header: buf_parse_u16_decimal");
      goto restore;
    }
    result += r;
    if (! str_init_alloc_copy(&tmp, result, buf->ptr.pchar + save.rpos)) {
      r = -1;
      goto clean;
    }
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_file (s_buf *buf, s_pdf_file *dest)
  {
    sw r;
    sw result = 0;
    s_pdf_file tmp = {0};
    if ((r = pdf_buf_parse_file_header(buf, &tmp.header)) <= 0) {
      err_puts("pdf_buf_parse_file: pdf_buf_parse_file_header");
      assert(! "pdf_buf_parse_file: pdf_buf_parse_file_header");
      return r;
    }
    result += r;
    if ((r = pdf_buf_parse_trailer(buf, &tmp.trailer) <= 0)) {
      err_puts("pdf_buf_parse_file: pdf_buf_parse_trailer");
      assert(! "pdf_buf_parse_file: pdf_buf_parse_trailer");
      return r;
    }
    result += r;
    if ((r = buf_seek(buf, tmp.trailer.startxref, SEEK_SET)) < 0 ||
        (u64) r != tmp.trailer.startxref) {
      err_puts("pdf_buf_parse_file: buf_seek");
      assert(! "pdf_buf_parse_file: buf_seek");
      return -1;
    }
    if ((r = pdf_buf_parse_xref(buf, &tmp.xref)) <= 0) {
      err_puts("pdf_buf_parse_file: pdf_buf_parse_xref");
      assert(! "pdf_buf_parse_file: pdf_buf_parse_xref");
      pdf_file_clean(&tmp);
      return -1;
    }
    result += r;
    *dest = tmp;
    return result;
  }
  
  sw pdf_buf_parse_float (s_buf *buf, f64 *dest)
  {
    u8 digit;
    bool end;
    f64 i;
    bool negative = false;
    sw r;
    sw result = 0;
    s_buf_save save;
    f64 tmp;
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, "+")) < 0)
      goto clean;
    if (r)
      result += r;
    else {
      if ((r = buf_read_1(buf, "-")) < 0)
        goto clean;
      if (r) {
        negative = true;
        result += r;
      }
    }
    if ((r = buf_parse_digit_dec(buf, &digit)) <= 0)
      goto restore;
    result += r;
    tmp = digit;
    while ((r = buf_parse_digit_dec(buf, &digit)) > 0) {
      result += r;
      tmp = tmp * 10 + digit;
    }
    if (r < 0 ||
        (r = buf_read_1(buf, ".")) <= 0)
      goto ok;
    result += r;
    i = 10;
    while ((r = buf_parse_digit_dec(buf, &digit)) > 0) {
      result += r;
      tmp += (f32) digit / i;
      i *= 10;
    }
    if ((r = pdf_buf_parse_object_end(buf, &end)) > 0)
      result += r;
    if (! end) {
      r = -1;
      goto restore;
    }
    result += r;
   ok:
    if (negative)
      tmp = -tmp;
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_indirect_object (s_buf *buf, s_tuple *dest)
  {
    s_tag generation_number = {0};
    s_tag object_number = {0};
    sw r;
    sw result = 0;
    s_buf_save save = {0};
    const s_sym *sym_indirect_object = sym_1("indirect_object");
    s_tuple tmp = {0};
    buf_save_init(buf, &save);
    if ((r = pdf_buf_parse_integer(buf, &object_number)) <= 0)
      goto clean;
    result += r;
    if ((r = pdf_buf_parse_integer(buf, &generation_number)) <= 0)
      goto restore;
    result += r;
    if ((r = pdf_buf_parse_token(buf, "obj")) < 0)
      goto restore;
    if (r) {
      result += r;
      if (! tuple_init(&tmp, 4)) {
        r = -1;
        goto restore;
      }
      if ((r = pdf_buf_parse_stream(buf, tmp.tag + 3)) < 0)
        goto restore;
      if (! r &&
          (r = pdf_buf_parse(buf, tmp.tag + 3)) <= 0)
        goto restore;
      goto ok;
    }
    if ((r = pdf_buf_parse_token(buf, "R")) < 0)
      goto restore;
    if (r) {
      result += r;
      if (! tuple_init(&tmp, 3)) {
        r = -1;
        goto restore;
      }
      goto ok;
    }
    goto restore;
   ok:
    tag_init_psym(tmp.tag, sym_indirect_object);
    tmp.tag[1] = object_number;
    tmp.tag[2] = generation_number;
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    tuple_clean(&tmp);
    tag_clean(&object_number);
    tag_clean(&generation_number);
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_integer (s_buf *buf, s_tag *dest)
  {
    bool end;
    bool negative = false;
    sw r;
    sw result = 0;
    s_buf_save save = {0};
    s_tag tmp = {0};
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, "+")) < 0)
      goto clean;
    if (! r) {
      if ((r = buf_read_1(buf, "-")) < 0)
        goto clean;
      if (r)
        negative = true;
    }
    result += r;
    if ((r = buf_parse_integer_decimal(buf, negative,
                                       &tmp.data.integer)) <= 0)
      goto restore;
    result += r;
    if ((r = pdf_buf_parse_object_end(buf, &end)) > 0)
      result += r;
    if (! end) {
      r = -1;
      goto restore;
    }
    tmp.type = TAG_INTEGER;
    tag_integer_reduce(&tmp);
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_name (s_buf *buf, p_sym *dest)
  {
    char a[BUF_SIZE];
    character c;
    u8 d;
    u8 d1;
    bool end;
    sw r;
    sw result = 0;
    s_buf_save save = {0};
    s_str str = {0};
    p_sym tmp = {0};
    s_buf tmp_buf;
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, "/")) <= 0)
      goto clean;
    result += r;
    buf_init(&tmp_buf, false, sizeof(a) - 1, a);
    while (1) {
      if ((r = pdf_buf_parse_object_end(buf, &end)) > 0)
        result += r;
      if (end)
        goto ok;
      if ((r = buf_read_character_utf8(buf, &c)) <= 0)
        break;
      result += r;
      if (c == '#') {
        if ((r = buf_read_character_utf8(buf, &c)) <= 0)
          break;
        result += r;
        if (c >= '0' && c <= '9')
          d = c - '0';
        else if (c >= 'A' && c <= 'F')
          d = c - 'A';
        else if (c >= 'a' && c <= 'f')
          d = c - 'a';
        else {
          r = 0;
          goto restore;
        }
        if ((r = buf_read_character_utf8(buf, &c)) <= 0)
          break;
        result += r;
        if (c >= '0' && c <= '9')
          d1 = c - '0';
        else if (c >= 'A' && c <= 'F')
          d1 = c - 'A';
        else if (c >= 'a' && c <= 'f')
          d1 = c - 'a';
        else {
          r = 0;
          goto restore;
        }
        c = d * 0x10 + d1;
      }
      if ((r = buf_write_character_utf8(&tmp_buf, c)) <= 0)
        goto restore;
    }
   ok:
    if (! buf_read_to_str(&tmp_buf, &str)) {
      r = -2;
      goto clean;
    }
    if (! (tmp = str_to_sym(&str))) {
      r = -2;
      str_clean(&str);
      goto clean;
    }
    str_clean(&str);
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_null (s_buf *buf, s_tag *dest)
  {
    sw r;
    if ((r = pdf_buf_parse_token(buf, "null")) > 0)
      tag_init_void(dest);
    return r;
  }
  
  sw pdf_buf_parse_number (s_buf *buf, s_tag *dest)
  {
    sw r;
    s_tag tmp = {0};
    assert(buf);
    assert(dest);
    if ((r = pdf_buf_parse_integer(buf, &tmp)) > 0)
      goto ok;
    if ((r = pdf_buf_parse_float(buf, &tmp.data.f64)) < 0) {
      tmp.type = TAG_F64;
      goto ok;
    }
    return 0;
   ok:
    *dest = tmp;
    return r;
  }
  
  sw pdf_buf_parse_object_end (s_buf *buf, bool *end)
  {
    character c;
    sw r;
    sw result = 0;
    if ((r = buf_ignore_spaces(buf)) < 0)
      goto end;
    result += r;
    if ((r = pdf_buf_parse_comments(buf)) < 0)
      goto end;
    result += r;
    if ((r = buf_peek_character_utf8(buf, &c)) < 0)
      goto end;
    if (r && pdf_character_is_delimiter(c))
      goto end;
    *end = result > 0 ? true : false;
    return result;
   end:
    *end = true;
    return result;
  }
  
  sw pdf_buf_parse_rewind_to_trailer (s_buf *buf)
  {
    bool end;
    sw pos;
    sw r;
    sw result = 0;
    assert(buf);
    if (false) {
      err_write_1("pdf_buf_parse_rewind_to_trailer: rpos = ");
      err_inspect_sw_decimal(buf->rpos);
      err_write_1("\n");
    }
    if (buf->rpos < 9)
      return -1;
    buf->rpos -= 9;
    while (1) {
      if (! buf->rpos)
        return -1;
      buf->rpos--;
      pos = buf->rpos;
      if ((r = pdf_buf_parse_object_end(buf, &end)) > 0 &&
          (result += r) &&
          end &&
          (r = pdf_buf_parse_token(buf, "trailer")) > 0) {
        result += r;
        return result;
      }
      buf->rpos = pos;
    }
    return -1;
  }
      
  sw pdf_buf_parse_stream (s_buf *buf, s_tag *dest)
  {
    s_pdf_stream *pdf_stream;
    s_map map = {0};
    sw r;
    sw result = 0;
    s_buf_save save;
    const s_sym *sym_PDF_Stream = sym_1("PDF.Stream");
    s_tag tmp = {0};
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = pdf_buf_parse_dictionnary(buf, &map)) <= 0)
      goto clean;
    result += r;
    if ((r = buf_read_1(buf, "stream")) <= 0) {
      map_clean(&map);
      goto restore;
    }
    result += r;
    if ((r = buf_read_1(buf, "\r")) < 0) {
      map_clean(&map);
      goto restore;
    }
    result += r;
    if ((r = buf_read_1(buf, "\n")) <= 0) {
      map_clean(&map);
      goto restore;
    }
    result += r;
    if (! tag_init_pstruct(&tmp, sym_PDF_Stream) ||
        ! struct_allocate(tmp.data.pstruct) ||
        ! (pdf_stream = tmp.data.pstruct->data)) {
      tag_clean(&tmp);
      map_clean(&map);
      r = -1;
      goto clean;
    }
    pdf_stream->dictionnary = map;
    if (! buf_tell_r(buf, &pdf_stream->offset)) {
      tag_clean(&tmp);
      r = -1;
      goto clean;
    }
    if ((r = pdf_buf_ignore_until_token(buf, "endstream")) <= 0) {
      tag_clean(&tmp);
      goto restore;
    }
    result += r;
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_string (s_buf *buf, s_tag *dest)
  {
    sw r;
    sw result = 0;
    s_buf_save save = {0};
    s_str str = {0};
    s_tag tmp = {0};
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = pdf_buf_parse_string_paren(buf, &str)) < 0)
      goto clean;
    if (r) {
      result += r;
      goto ok;
    }
    if ((r = pdf_buf_parse_string_hex(buf, &str)) < 0)
      goto clean;
    if (r) {
      result += r;
      goto ok;
    }
    r = 0;
    goto clean;
   ok:
    if (false) {
      err_write_1("pdf_buf_parse_string: str = ");
      err_inspect_str(&str);
      err_write_1("\n");
    }
    // TODO: fix
    if (false && ! str_parse_eval(&str, &tmp)) {
      err_puts("pdf_buf_parse_string: str_parse_eval");
      assert(! "pdf_buf_parse_string: str_parse_eval");
      r = -1;
      str_clean(&str);
      goto restore;
    }
    //
    tmp.type = TAG_STR;
    tmp.data.str = str;
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_string_hex (s_buf *buf, s_str *dest)
  {
    char a[BUF_SIZE] = {0};
    character c;
    u8 d;
    u8 d1;
    bool end;
    sw r;
    sw result = 0;
    s_buf_save save = {0};
    s_str tmp = {0};
    s_buf tmp_buf = {0};
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, "<")) <= 0)
      goto clean;
    result += r;
    buf_init(&tmp_buf, false, sizeof(a) - 1, a);
    while (1) {
      if ((r = buf_read_character_utf8(buf, &c)) <= 0)
        goto restore;
      result += r;
      if (c == '>')
        break;
      if (c >= '0' && c <= '9')
        d = c - '0';
      else if (c >= 'a' && c <= 'f')
        d = c - 'a' + 10;
      else if (c >= 'A' && c <= 'F')
        d = c - 'A' + 10;
      else {
        r = -1;
        goto restore;
      }
      if ((r = buf_read_character_utf8(buf, &c)) <= 0)
        goto restore;
      result += r;
      if (c >= '0' && c <= '9')
        d1 = c - '0';
      else if (c >= 'a' && c <= 'f')
        d1 = c - 'a' + 10;
      else if (c >= 'A' && c <= 'F')
        d1 = c - 'A' + 10;
      else {
        r = -1;
        goto restore;
      }
      if ((r = buf_write_u8(&tmp_buf, (d << 4) | d1)) <= 0)
        goto restore;
    }
    if ((r = pdf_buf_parse_object_end(buf, &end)) > 0)
      result += r;
    if (! end) {
      r = -1;
      goto restore;
    }
    if (! buf_read_to_str(&tmp_buf, &tmp)) {
      r = -1;
      goto restore;
    }
    buf_clean(&tmp_buf);
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_clean(&tmp_buf);
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  // parentheses must be balanced or backslashed
  sw pdf_buf_parse_string_paren (s_buf *buf, s_str *dest)
  {
    char a[BUF_SIZE] = {0};
    character c;
    character c1;
    character c2;
    s32 paren = 0;
    sw r;
    sw result = 0;
    s_buf_save save = {0};
    s_str tmp = {0};
    s_buf tmp_buf = {0};
    assert(buf);
    assert(dest);
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, "(")) <= 0)
      goto clean;
    result += r;
    paren++;
    buf_init(&tmp_buf, false, sizeof(a) - 1, a);
    while (1) {
      if ((r = buf_read_character_utf8(buf, &c)) <= 0)
        goto restore;
      result += r;
      if (c == '(')
        paren++;
      else if (c == ')') {
        paren--;
        if (! paren)
          goto ok;
      }
      else if (c == '\\') {
        if ((r = buf_read_character_utf8(buf, &c)) <= 0)
          goto restore;
        result += r;
        switch (c) {
        case 'n':  c = '\n'; break;
        case 'r':  c = '\r'; break;
        case 't':  c = '\t'; break;
        case 'b':  c = '\b'; break;
        case 'f':  c = '\f'; break;
        case '(':  c = '(';  break;
        case ')':  c = ')';  break;
        case '\\': c = '\\'; break;
        case '0': case '1': case '2': case '3':
          if ((r = buf_read_character_utf8(buf, &c1)) <= 0)
            break;
          if (c1 < '0' || c1 > '7') {
            r = -1;
            break;
          }
          result += r;
          if ((r = buf_read_character_utf8(buf, &c2)) <= 0)
            break;
          if (c2 < '0' || c2 > '7') {
            r = -1;
            break;
          }
          result += r;
          c = (c - '0') * 0100 + (c1 - '0') * 010 + (c2 - '0');
          break;
        case '\n':
          goto next;
        }
      }
      if ((r = buf_write_character_utf8(&tmp_buf, c)) <= 0)
        goto restore;
    next:
      continue;
    }
   ok:
    if ((r = buf_read_to_str(&tmp_buf, &tmp)) < 0)
      goto restore;
    *dest = tmp;
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_token (s_buf *buf, const char *pchar)
  {
    bool end;
    sw r;
    sw result = 0;
    s_buf_save save;
    buf_save_init(buf, &save);
    if ((r = buf_read_1(buf, pchar)) <= 0)
      goto clean;
    result += r;
    if ((r = pdf_buf_parse_object_end(buf, &end)) < 0)
      goto restore;
    if (! end) {
      r = 0;
      goto restore;
    }
    r = result;
    goto clean;
   restore:
    buf_save_restore_rpos(buf, &save);
   clean:
    buf_save_clean(buf, &save);
    return r;
  }
  
  sw pdf_buf_parse_trailer (s_buf *buf, s_pdf_trailer *dest)
  {
    u64 pos;
    sw r;
    sw result = 0;
    sw size;
    const s_sym *sym_U64 = &g_sym_U64;
    s_tag tag = {0};
    s_pdf_trailer tmp = {0};
    if (! buf_total_size(buf, &pos) || ! pos) {
      err_puts("pdf_buf_parse_trailer: invalid buf_total_size");
      assert(! "pdf_buf_parse_trailer: invalid buf_total_size");
      return -1;
    }
    if (pos < buf->size) {
      size = pos;
      pos = 0;
    }
    else {
      size = buf->size;
      pos -= size;
    }
    if ((r = buf_seek(buf, pos, SEEK_SET)) < 0 ||
        (uw) r != pos) {
      err_puts("pdf_buf_parse_trailer: buf_seek");
      assert(! "pdf_buf_parse_trailer: buf_seek");
      return r;
    }
    if (buf_refill(buf, size) < size) {
      err_puts("pdf_buf_parse_trailer: buf_refill");
      assert(! "pdf_buf_parse_trailer: buf_refill");
      return -1;
    }
    buf->rpos = buf->wpos;
    if ((r = pdf_buf_parse_rewind_to_trailer(buf)) <= 0) {
      err_puts("pdf_buf_parse_trailer: pdf_buf_parse_rewind_to_trailer");
      assert(! "pdf_buf_parse_trailer: pdf_buf_parse_rewind_to_trailer");
      return r;
    }
    result += r;
    if ((r = pdf_buf_parse_dictionnary(buf, &tmp.dictionnary)) <= 0) {
      err_puts("pdf_buf_parse_trailer: pdf_buf_parse_dictionnary");
      err_inspect_buf(buf);
      assert(! "pdf_buf_parse_trailer: pdf_buf_parse_dictionnary");
      return r;
    }
    result += r;
    if ((r = pdf_buf_parse_token(buf, "startxref")) <= 0) {
      err_puts("pdf_buf_parse_trailer: pdf_buf_parse_token startxref");
      assert(! "pdf_buf_parse_trailer: pdf_buf_parse_token startxref");
      return r;
    }
    result += r;
    if ((r = pdf_buf_parse_integer(buf, &tag)) <= 0) {
      err_puts("pdf_buf_parse_trailer: pdf_buf_parse_integer");
      assert(! "pdf_buf_parse_trailer: pdf_buf_parse_integer");
      return r;
    }
    result += r;
    if (! u64_init_cast(&tmp.startxref, &sym_U64, &tag)) {
      err_puts("pdf_buf_parse_trailer: u64_init_cast");
      assert(! "pdf_buf_parse_trailer: u64_init_cast");
      tag_clean(&tag);
      return -1;
    }
    if (false) {
      err_write_1("pdf_buf_parse_trailer: startxref = ");
      err_inspect_u64_decimal(tmp.startxref);
      err_write_1("\n");
    }
    tag_clean(&tag);
    if (buf->rpos < buf->wpos) {
      err_puts("pdf_buf_parse_trailer: data after %%EOF");
      assert(! "pdf_buf_parse_trailer: data after %%EOF");
      return -1;
    }
    result += r;
    *dest = tmp;
    return result;
  }
  
  sw pdf_buf_parse_xref (s_buf *buf, s_map *dest)
  {
    char c;
    u32 count;
    u16 generation_number;
    u32 i;
    p_list  keys;
    p_list *keys_tail;
    u32 object_number;
    u32 offset;
    sw r;
    sw r1;
    sw r2;
    sw r3;
    sw r4;
    sw r5;
    sw result = 0;
    sw result_inner;
    s_buf_save save;
    s_map tmp = {0};
    p_list  values;
    p_list *values_tail;
    assert(buf);
    assert(dest);
    if ((r = pdf_buf_parse_token(buf, "xref")) <= 0) {
      err_puts(" pdf_buf_parse_xref: pdf_buf_parse_token");
      assert(! " pdf_buf_parse_xref: pdf_buf_parse_token");
      return r;
    }
    result += r;
    buf_save_init(buf, &save);
    result_inner = result;
    keys = NULL;
    keys_tail = &keys;
    values = NULL;
    values_tail = &values;
    while (1) {
      buf_save_update(buf, &save);
      result = result_inner;
      if ((r = buf_parse_u32_decimal(buf, &object_number)) <= 0 ||
          (r1 = buf_read_1(buf, " ")) <= 0 ||
          (r2 = buf_parse_u32_decimal(buf, &count)) <= 0 ||
          ((r3 = buf_read_1(buf, "\n")) <= 0 &&
           (r3 = buf_read_1(buf, "\r\n")) <= 0)) {
        break;
      }
      result_inner += r + r1 + r2 + r3;
      i = 0;
      while (i < count) {
        if ((r = buf_parse_u32_decimal(buf, &offset)) != 10 ||
            (r1 = buf_read_1(buf, " ")) != 1 ||
            (r2 = buf_parse_u16_decimal(buf, &generation_number)) != 5 ||
            (r3 = buf_read_1(buf, " ")) != 1 ||
            (r4 = buf_read_u8(buf, (u8 *) &c)) != 1 ||
            ((r5 = buf_read_1(buf, " \n")) <= 0 &&
             (r5 = buf_read_1(buf, "\r\n")) != 2)) {
          err_puts("pdf_buf_parse_xref: invalid xref");
          err_inspect_buf(buf);
          assert(! "pdf_buf_parse_xref: invalid xref");
          goto error;
        }
        result_inner += r + r1 + r2 + r3 + r4 + r5;
        if (! (*keys_tail = list_new_tuple(2, NULL)) ||
            ! (*values_tail = list_new_u64(offset, NULL)))
          goto error;
        tag_init_u32((*keys_tail)->tag.data.tuple.tag, object_number);
        tag_init_u16((*keys_tail)->tag.data.tuple.tag + 1, generation_number);
        keys_tail = &(*keys_tail)->next.data.plist;
        values_tail = &(*values_tail)->next.data.plist;
        i++;
        object_number++;
      }
    }
    //restore_inner:
    buf_save_restore_rpos(buf, &save);
    buf_save_clean(buf, &save);
    if (! keys || ! values) {
      err_puts("pdf_buf_parse_xref: empty xref");
      err_inspect_buf(buf);
      assert(! "pdf_buf_parse_xref: empty xref");
      return -1;
    }
    if (! map_init_from_lists(&tmp, keys, values)) {
      err_puts("pdf_buf_parse_xref: map_init_from_lists");
      assert(! "pdf_buf_parse_xref: map_init_from_lists");
      list_delete_all(keys);
      list_delete_all(values);
      return -1;
    }
    *dest = tmp;
    return result;
   error:
    list_delete_all(keys);
    list_delete_all(values);
    buf_save_restore_rpos(buf, &save);
    buf_save_clean(buf, &save);
    return -1;
  }
  
  bool pdf_character_is_delimiter (character c)
  {
    return (c == '%' ||
            c == '/' ||
            c == '>' ||
            c == ']' ||
            character_is_space(c));
  }
  

Download