 |
remove unneeded code, handle - grabtitle - stupid HTML title grabber |
 |
git clone git://git.codemadness.org/grabtitle |
 |
Log |
|
Files |
|
Refs |
|
README |
|
LICENSE |
|
--- |
|
commit 16cc59c155068e6de1fd5cfa8720d6d765db6548 |
|
parent 16dfed456fd96d1c483eb515594019d7a5febc86 |
|
Author: Hiltjo Posthuma <[email protected]> |
|
Date: Sun, 22 Sep 2019 19:50:21 +0200 |
|
|
|
remove unneeded code, handle |
|
|
|
Diffstat: |
|
M xml.c | 136 ++++-------------------------… |
|
M xml.h | 19 ------------------- |
|
|
|
2 files changed, 18 insertions(+), 137 deletions(-) |
|
--- |
|
diff --git a/xml.c b/xml.c |
|
@@ -12,7 +12,7 @@ |
|
static void |
|
xml_parseattrs(XMLParser *x) |
|
{ |
|
- size_t namelen = 0, valuelen; |
|
+ size_t namelen = 0; |
|
int c, endsep, endname = 0, valuestart = 0; |
|
|
|
while ((c = x->getnext()) != EOF) { |
|
@@ -23,94 +23,33 @@ xml_parseattrs(XMLParser *x) |
|
} else if (c == '?') |
|
; /* ignore */ |
|
else if (c == '=') { |
|
- x->name[namelen] = '\0'; |
|
valuestart = 1; |
|
endname = 1; |
|
} else if (namelen && ((endname && !valuestart && isalpha(c)) … |
|
- /* attribute without value */ |
|
- x->name[namelen] = '\0'; |
|
- if (x->xmlattrstart) |
|
- x->xmlattrstart(x, x->tag, x->taglen, x->name,… |
|
- if (x->xmlattr) |
|
- x->xmlattr(x, x->tag, x->taglen, x->name, name… |
|
- if (x->xmlattrend) |
|
- x->xmlattrend(x, x->tag, x->taglen, x->name, n… |
|
endname = 0; |
|
- x->name[0] = c; |
|
namelen = 1; |
|
} else if (namelen && valuestart) { |
|
/* attribute with value */ |
|
- if (x->xmlattrstart) |
|
- x->xmlattrstart(x, x->tag, x->taglen, x->name,… |
|
- |
|
- valuelen = 0; |
|
if (c == '\'' || c == '"') { |
|
endsep = c; |
|
- } else { |
|
- endsep = ' '; /* isspace() */ |
|
- goto startvalue; |
|
- } |
|
- |
|
- while ((c = x->getnext()) != EOF) { |
|
-startvalue: |
|
- if (c == '&') { /* entities */ |
|
- x->data[valuelen] = '\0'; |
|
- /* call data function with data before… |
|
- if (valuelen && x->xmlattr) |
|
- x->xmlattr(x, x->tag, x->tagle… |
|
- x->data[0] = c; |
|
- valuelen = 1; |
|
- while ((c = x->getnext()) != EOF) { |
|
- if (c == endsep || (endsep == … |
|
- break; |
|
- if (valuelen < sizeof(x->data)… |
|
- x->data[valuelen++] = … |
|
- else { |
|
- /* entity too long for… |
|
- x->data[valuelen] = '\… |
|
- if (x->xmlattr) |
|
- x->xmlattr(x, … |
|
- x->data[0] = c; |
|
- valuelen = 1; |
|
- break; |
|
- } |
|
- if (c == ';') { |
|
- x->data[valuelen] = '\… |
|
- if (x->xmlattrentity) |
|
- x->xmlattrenti… |
|
- valuelen = 0; |
|
- break; |
|
- } |
|
- } |
|
- } else if (c != endsep && !(endsep == ' ' && (… |
|
- if (valuelen < sizeof(x->data) - 1) { |
|
- x->data[valuelen++] = c; |
|
- } else { |
|
- x->data[valuelen] = '\0'; |
|
- if (x->xmlattr) |
|
- x->xmlattr(x, x->tag, … |
|
- x->data[0] = c; |
|
- valuelen = 1; |
|
- } |
|
+ while ((c = x->getnext()) != EOF) { |
|
+ if (c == endsep) |
|
+ break; |
|
} |
|
- if (c == endsep || (endsep == ' ' && (c == '>'… |
|
- x->data[valuelen] = '\0'; |
|
- if (x->xmlattr) |
|
- x->xmlattr(x, x->tag, x->tagle… |
|
- if (x->xmlattrend) |
|
- x->xmlattrend(x, x->tag, x->ta… |
|
- break; |
|
+ } else { |
|
+ while ((c = x->getnext()) != EOF) { |
|
+ if (c == '>' || isspace(c)) |
|
+ break; |
|
} |
|
} |
|
namelen = endname = valuestart = 0; |
|
- } else if (namelen < sizeof(x->name) - 1) { |
|
- x->name[namelen++] = c; |
|
+ } else { |
|
+ namelen = 1; |
|
} |
|
if (c == '>') { |
|
break; |
|
} else if (c == '/') { |
|
x->isshorttag = 1; |
|
- x->name[0] = '\0'; |
|
namelen = 0; |
|
} |
|
} |
|
@@ -119,48 +58,17 @@ startvalue: |
|
static void |
|
xml_parsecomment(XMLParser *x) |
|
{ |
|
- size_t datalen = 0, i = 0; |
|
+ size_t i = 0; |
|
int c; |
|
|
|
- if (x->xmlcommentstart) |
|
- x->xmlcommentstart(x); |
|
while ((c = x->getnext()) != EOF) { |
|
- if (c == '-' || c == '>') { |
|
- if (x->xmlcomment) { |
|
- x->data[datalen] = '\0'; |
|
- x->xmlcomment(x, x->data, datalen); |
|
- datalen = 0; |
|
- } |
|
- } |
|
- |
|
if (c == '-') { |
|
- if (++i > 2) { |
|
- if (x->xmlcomment) |
|
- for (; i > 2; i--) |
|
- x->xmlcomment(x, "-", 1); |
|
- i = 2; |
|
- } |
|
- continue; |
|
+ if (i < 2) |
|
+ i++; |
|
} else if (c == '>' && i == 2) { |
|
- if (x->xmlcommentend) |
|
- x->xmlcommentend(x); |
|
return; |
|
- } else if (i) { |
|
- if (x->xmlcomment) { |
|
- for (; i > 0; i--) |
|
- x->xmlcomment(x, "-", 1); |
|
- } |
|
- i = 0; |
|
- } |
|
- |
|
- if (datalen < sizeof(x->data) - 1) { |
|
- x->data[datalen++] = c; |
|
} else { |
|
- x->data[datalen] = '\0'; |
|
- if (x->xmlcomment) |
|
- x->xmlcomment(x, x->data, datalen); |
|
- x->data[0] = c; |
|
- datalen = 1; |
|
+ i = 0; |
|
} |
|
} |
|
} |
|
@@ -171,8 +79,6 @@ xml_parsecdata(XMLParser *x) |
|
size_t datalen = 0, i = 0; |
|
int c; |
|
|
|
- if (x->xmlcdatastart) |
|
- x->xmlcdatastart(x); |
|
while ((c = x->getnext()) != EOF) { |
|
if (c == ']' || c == '>') { |
|
if (x->xmlcdata) { |
|
@@ -191,10 +97,8 @@ xml_parsecdata(XMLParser *x) |
|
} |
|
continue; |
|
} else if (c == '>' && i == 2) { |
|
- if (x->xmlcdataend) |
|
- x->xmlcdataend(x); |
|
return; |
|
- } else if (i) { |
|
+ } else { |
|
if (x->xmlcdata) |
|
for (; i > 0; i--) |
|
x->xmlcdata(x, "]", 1); |
|
@@ -255,11 +159,13 @@ namedentitytostr(const char *e, char *buf, size_t bufsiz) |
|
{ ">", '>' }, |
|
{ "'", '\'' }, |
|
{ """, '"' }, |
|
+ { " ", ' ' }, |
|
{ "&", '&' }, |
|
{ "<", '<' }, |
|
{ ">", '>' }, |
|
{ "&APOS;", '\'' }, |
|
- { """, '"' } |
|
+ { """, '"' }, |
|
+ { "&NBSP;", ' ' }, |
|
}; |
|
size_t i; |
|
|
|
@@ -398,8 +304,6 @@ xml_parse(XMLParser *x) |
|
x->xmltagstart… |
|
if (isspace(c)) |
|
xml_parseattrs… |
|
- if (x->xmltagstartpars… |
|
- x->xmltagstart… |
|
} |
|
/* call tagend for shortform o… |
|
if (x->isshorttag) { |
|
@@ -416,8 +320,6 @@ xml_parse(XMLParser *x) |
|
} else { |
|
/* parse tag data */ |
|
datalen = 0; |
|
- if (x->xmldatastart) |
|
- x->xmldatastart(x); |
|
while ((c = x->getnext()) != EOF) { |
|
if (c == '&') { |
|
if (datalen) { |
|
@@ -464,8 +366,6 @@ xml_parse(XMLParser *x) |
|
x->data[datalen] = '\0'; |
|
if (x->xmldata && datalen) |
|
x->xmldata(x, x->data, datalen… |
|
- if (x->xmldataend) |
|
- x->xmldataend(x); |
|
break; |
|
} |
|
} |
|
diff --git a/xml.h b/xml.h |
|
@@ -1,27 +1,10 @@ |
|
typedef struct xmlparser { |
|
/* handlers */ |
|
- void (*xmlattr)(struct xmlparser *, const char *, size_t, |
|
- const char *, size_t, const char *, size_t); |
|
- void (*xmlattrend)(struct xmlparser *, const char *, size_t, |
|
- const char *, size_t); |
|
- void (*xmlattrstart)(struct xmlparser *, const char *, size_t, |
|
- const char *, size_t); |
|
- void (*xmlattrentity)(struct xmlparser *, const char *, size_t, |
|
- const char *, size_t, const char *, size_t); |
|
- void (*xmlcdatastart)(struct xmlparser *); |
|
void (*xmlcdata)(struct xmlparser *, const char *, size_t); |
|
- void (*xmlcdataend)(struct xmlparser *); |
|
- void (*xmlcommentstart)(struct xmlparser *); |
|
- void (*xmlcomment)(struct xmlparser *, const char *, size_t); |
|
- void (*xmlcommentend)(struct xmlparser *); |
|
void (*xmldata)(struct xmlparser *, const char *, size_t); |
|
- void (*xmldataend)(struct xmlparser *); |
|
void (*xmldataentity)(struct xmlparser *, const char *, size_t); |
|
- void (*xmldatastart)(struct xmlparser *); |
|
void (*xmltagend)(struct xmlparser *, const char *, size_t, int); |
|
void (*xmltagstart)(struct xmlparser *, const char *, size_t); |
|
- void (*xmltagstartparsed)(struct xmlparser *, const char *, |
|
- size_t, int); |
|
|
|
int (*getnext)(void); |
|
|
|
@@ -30,8 +13,6 @@ typedef struct xmlparser { |
|
size_t taglen; |
|
/* current tag is in short form ? <tag /> */ |
|
int isshorttag; |
|
- /* current attribute name */ |
|
- char name[1024]; |
|
/* data buffer used for tag data, cdata and attribute data */ |
|
char data[BUFSIZ]; |
|
} XMLParser; |
