diff options
Diffstat (limited to 'src/scanners.re')
| -rw-r--r-- | src/scanners.re | 85 |
1 files changed, 34 insertions, 51 deletions
diff --git a/src/scanners.re b/src/scanners.re index 7323ef9..5ac7c15 100644 --- a/src/scanners.re +++ b/src/scanners.re @@ -1,8 +1,15 @@ -#include "buffer.h" +#include "scanners.h" + +#define SCAN_DATA \ + const unsigned char *marker = NULL; \ + const unsigned char *p = c->data + offset; \ + const unsigned char *start = p; \ + const unsigned char *end = c->data + c->len /*!re2c re2c:define:YYCTYPE = "unsigned char"; re2c:define:YYCURSOR = p; + re2c:define:YYLIMIT = end; re2c:define:YYMARKER = marker; re2c:define:YYCTXMARKER = marker; re2c:yyfill:enable = 0; @@ -55,11 +62,9 @@ */ // Try to match URI autolink after first <, returning number of chars matched. -extern int scan_autolink_uri(const gh_buf *s, int pos) +extern int scan_autolink_uri(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c scheme [:]([^\x00-\x20<>\\]|escaped_char)*[>] { return (p - start); } .? { return 0; } @@ -67,11 +72,9 @@ extern int scan_autolink_uri(const gh_buf *s, int pos) } // Try to match email autolink after first <, returning num of chars matched. -extern int scan_autolink_email(const gh_buf *s, int pos) +extern int scan_autolink_email(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c [a-zA-Z0-9.!#$%&'*+/=?^_`{|}~-]+ [@] @@ -83,11 +86,9 @@ extern int scan_autolink_email(const gh_buf *s, int pos) } // Try to match an HTML tag after first <, returning num of chars matched. -extern int scan_html_tag(const gh_buf *s, int pos) +extern int scan_html_tag(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c htmltag { return (p - start); } .? { return 0; } @@ -96,11 +97,9 @@ extern int scan_html_tag(const gh_buf *s, int pos) // Try to match an HTML block tag including first <, // returning num of chars matched. -extern int scan_html_block_tag(const gh_buf *s, int pos) +extern int scan_html_block_tag(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c [<] [/] blocktagname (spacechar | [>]) { return (p - start); } [<] blocktagname (spacechar | [/>]) { return (p - start); } @@ -113,11 +112,9 @@ extern int scan_html_block_tag(const gh_buf *s, int pos) // This may optionally be contained in <..>; otherwise // whitespace and unbalanced right parentheses aren't allowed. // Newlines aren't ever allowed. -extern int scan_link_url(const gh_buf *s, int pos) +extern int scan_link_url(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c [ \n]* [<] ([^<>\n\\\x00] | escaped_char | [\\])* [>] { return (p - start); } [ \n]* (reg_char+ | escaped_char | in_parens_nosp)* { return (p - start); } @@ -128,11 +125,9 @@ extern int scan_link_url(const gh_buf *s, int pos) // Try to match a link title (in single quotes, in double quotes, or // in parentheses), returning number of chars matched. Allow one // level of internal nesting (quotes within quotes). -extern int scan_link_title(const gh_buf *s, int pos) +extern int scan_link_title(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c ["] (escaped_char|[^"\x00])* ["] { return (p - start); } ['] (escaped_char|[^'\x00])* ['] { return (p - start); } @@ -142,10 +137,9 @@ extern int scan_link_title(const gh_buf *s, int pos) } // Match space characters, including newlines. -extern int scan_spacechars(const gh_buf *s, int pos) +extern int scan_spacechars(chunk *c, int offset) { - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c [ \t\n]* { return (p - start); } . { return 0; } @@ -153,11 +147,9 @@ extern int scan_spacechars(const gh_buf *s, int pos) } // Match ATX header start. -extern int scan_atx_header_start(const gh_buf *s, int pos) +extern int scan_atx_header_start(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c [#]{1,6} ([ ]+|[\n]) { return (p - start); } .? { return 0; } @@ -166,10 +158,9 @@ extern int scan_atx_header_start(const gh_buf *s, int pos) // Match sexext header line. Return 1 for level-1 header, // 2 for level-2, 0 for no match. -extern int scan_setext_header_line(const gh_buf *s, int pos) +extern int scan_setext_header_line(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); + SCAN_DATA; /*!re2c [=]+ [ ]* [\n] { return 1; } [-]+ [ ]* [\n] { return 2; } @@ -180,11 +171,9 @@ extern int scan_setext_header_line(const gh_buf *s, int pos) // Scan a horizontal rule line: "...three or more hyphens, asterisks, // or underscores on a line by themselves. If you wish, you may use // spaces between the hyphens or asterisks." -extern int scan_hrule(const gh_buf *s, int pos) +extern int scan_hrule(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c ([*][ ]*){3,} [ \t]* [\n] { return (p - start); } ([_][ ]*){3,} [ \t]* [\n] { return (p - start); } @@ -194,11 +183,9 @@ extern int scan_hrule(const gh_buf *s, int pos) } // Scan an opening code fence. -extern int scan_open_code_fence(const gh_buf *s, int pos) +extern int scan_open_code_fence(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c [`]{3,} / [^`\n\x00]*[\n] { return (p - start); } [~]{3,} / [^~\n\x00]*[\n] { return (p - start); } @@ -207,11 +194,9 @@ extern int scan_open_code_fence(const gh_buf *s, int pos) } // Scan a closing code fence with length at least len. -extern int scan_close_code_fence(const gh_buf *s, int pos, int len) +extern int scan_close_code_fence(chunk *c, int offset, int len) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c ([`]{3,} | [~]{3,}) / spacechar* [\n] { if (p - start > len) { @@ -225,11 +210,9 @@ extern int scan_close_code_fence(const gh_buf *s, int pos, int len) // Scans an entity. // Returns number of chars matched. -extern int scan_entity(const gh_buf *s, int pos) +extern int scan_entity(chunk *c, int offset) { - unsigned char * marker = NULL; - unsigned char * p = &(s->ptr[pos]); - unsigned char * start = p; + SCAN_DATA; /*!re2c [&] ([#] ([Xx][A-Fa-f0-9]{1,8}|[0-9]{1,8}) |[A-Za-z][A-Za-z0-9]{1,31} ) [;] { return (p - start); } |