diff options
Diffstat (limited to 'src/xine-utils/xmlparser.c')
| -rw-r--r-- | src/xine-utils/xmlparser.c | 308 |
1 files changed, 260 insertions, 48 deletions
diff --git a/src/xine-utils/xmlparser.c b/src/xine-utils/xmlparser.c index a5d8212d2..a7bc146a9 100644 --- a/src/xine-utils/xmlparser.c +++ b/src/xine-utils/xmlparser.c @@ -19,6 +19,10 @@ * Floor, Boston, MA 02110, USA */ +#ifdef XINE_COMPILE +# include "config.h" +#endif + #include <unistd.h> #include <stdio.h> #include <stdlib.h> @@ -73,8 +77,11 @@ static xml_node_t * new_xml_node(void) { return new_node; } +static const char cdata[] = CDATA_MARKER; + static void free_xml_node(xml_node_t * node) { - free (node->name); + if (node->name != cdata) + free (node->name); free (node->data); free(node); } @@ -149,18 +156,79 @@ void xml_parser_free_tree(xml_node_t *current_node) { xml_parser_free_tree_rec(current_node, 1); } -#define STATE_IDLE 0 -#define STATE_NODE 1 -#define STATE_COMMENT 7 +typedef enum { + /*0*/ + STATE_IDLE, + /* <foo ...> */ + STATE_NODE, + STATE_ATTRIBUTE, + STATE_NODE_CLOSE, + STATE_TAG_TERM, + STATE_ATTRIBUTE_EQUALS, + STATE_STRING, + STATE_TAG_TERM_IGNORE, + /* <?foo ...?> */ + STATE_Q_NODE, + STATE_Q_ATTRIBUTE, + STATE_Q_NODE_CLOSE, + STATE_Q_TAG_TERM, + STATE_Q_ATTRIBUTE_EQUALS, + STATE_Q_STRING, + /* Others */ + STATE_COMMENT, + STATE_DOCTYPE, + STATE_CDATA, +} parser_state_t; + +static xml_node_t *xml_parser_append_text (xml_node_t *node, xml_node_t *subnode, const char *text, int flags) +{ + if (!text || !*text) + return subnode; /* empty string -> nothing to do */ + + if ((flags & XML_PARSER_MULTI_TEXT) && subnode) { + /* we have a subtree, so we can't use node->data */ + if (subnode->name == cdata) { + /* most recent node is CDATA - append to it */ + char *newtext; + asprintf (&newtext, "%s%s", subnode->data, text); + free (subnode->data); + subnode->data = newtext; + } else { + /* most recent node is not CDATA - add a sibling */ + subnode->next = new_xml_node (); + subnode->next->name = cdata; + subnode->next->data = strdup (text); + subnode = subnode->next; + } + } else if (node->data) { + /* "no" subtree, but we have existing text - append to it */ + char *newtext; + asprintf (&newtext, "%s%s", node->data, text); + free (node->data); + node->data = newtext; + } else { + /* no text, "no" subtree - duplicate & assign */ + while (isspace (*text)) + ++text; + if (*text) + node->data = strdup (text); + } + + return subnode; +} + +#define Q_STATE(CURRENT,NEW) (STATE_##NEW + state - STATE_##CURRENT) -static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int rec) { +static int xml_parser_get_node_internal (xml_node_t *current_node, char *root_names[], int rec, int flags) +{ char tok[TOKEN_SIZE]; char property_name[TOKEN_SIZE]; char node_name[TOKEN_SIZE]; - int state = STATE_IDLE; + parser_state_t state = STATE_IDLE; int res = 0; int parse_res; int bypass_get_token = 0; + int retval = 0; /* used when state==4; non-0 if there are missing </...> */ xml_node_t *subtree = NULL; xml_node_t *current_subtree = NULL; xml_property_t *current_property = NULL; @@ -182,30 +250,33 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r /* do nothing */ break; case (T_EOF): - return 0; /* normal end */ + return retval; /* normal end */ break; case (T_M_START_1): state = STATE_NODE; break; case (T_M_START_2): - state = 3; + state = STATE_NODE_CLOSE; break; case (T_C_START): state = STATE_COMMENT; break; case (T_TI_START): - state = 8; + state = STATE_Q_NODE; break; case (T_DOCTYPE_START): - state = 9; + state = STATE_DOCTYPE; + break; + case (T_CDATA_START): + state = STATE_CDATA; break; case (T_DATA): /* current data */ - if (current_node->data) { - /* avoid a memory leak */ - free(current_node->data); + { + char *decoded = lexer_decode_entities (tok); + current_subtree = xml_parser_append_text (current_node, current_subtree, decoded, flags); + free (decoded); } - current_node->data = lexer_decode_entities(tok); lprintf("info: node data : %s\n", current_node->data); break; default: @@ -216,6 +287,7 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r break; case STATE_NODE: + case STATE_Q_NODE: switch (res) { case (T_IDENT): properties = NULL; @@ -225,8 +297,13 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r if (xml_parser_mode == XML_PARSER_CASE_INSENSITIVE) { strtoupper(tok); } - strcpy(node_name, tok); - state = 2; + if (state == STATE_Q_NODE) { + snprintf (node_name, TOKEN_SIZE, "?%s", tok); + state = STATE_Q_ATTRIBUTE; + } else { + strcpy(node_name, tok); + state = STATE_ATTRIBUTE; + } lprintf("info: current node name \"%s\"\n", node_name); break; default: @@ -235,7 +312,8 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r break; } break; - case 2: + + case STATE_ATTRIBUTE: switch (res) { case (T_EOL): case (T_SEPAR): @@ -251,8 +329,9 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r /* set node propertys */ subtree->props = properties; lprintf("info: rec %d new subtree %s\n", rec, node_name); - parse_res = xml_parser_get_node(subtree, node_name, rec + 1); - if (parse_res != 0) { + root_names[rec + 1] = node_name; + parse_res = xml_parser_get_node_internal(subtree, root_names, rec + 1, flags); + if (parse_res == -1 || parse_res > 0) { return parse_res; } if (current_subtree == NULL) { @@ -262,11 +341,16 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r current_subtree->next = subtree; current_subtree = subtree; } + if (parse_res < -1) { + /* badly-formed XML (missing close tag) */ + return parse_res + 1 + (parse_res == -2); + } state = STATE_IDLE; break; case (T_M_STOP_2): /* new leaf */ /* new subtree */ + new_leaf: subtree = new_xml_node(); /* set node name */ @@ -288,11 +372,12 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r break; case (T_IDENT): /* save property name */ + new_prop: if (xml_parser_mode == XML_PARSER_CASE_INSENSITIVE) { strtoupper(tok); } strcpy(property_name, tok); - state = 5; + state = Q_STATE(ATTRIBUTE, ATTRIBUTE_EQUALS); lprintf("info: current property name \"%s\"\n", property_name); break; default: @@ -302,17 +387,50 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r } break; - case 3: + case STATE_Q_ATTRIBUTE: + switch (res) { + case (T_EOL): + case (T_SEPAR): + /* nothing */ + break; + case (T_TI_STOP): + goto new_leaf; + case (T_IDENT): + goto new_prop; + default: + lprintf("error: unexpected token \"%s\", state %d\n", tok, state); + return -1; + break; + } + break; + + case STATE_NODE_CLOSE: switch (res) { case (T_IDENT): /* must be equal to root_name */ if (xml_parser_mode == XML_PARSER_CASE_INSENSITIVE) { strtoupper(tok); } - if (strcmp(tok, root_name) == 0) { - state = 4; - } else { - lprintf("error: xml struct, tok=%s, waited_tok=%s\n", tok, root_name); + if (strcmp(tok, root_names[rec]) == 0) { + state = STATE_TAG_TERM; + } else if (flags & XML_PARSER_RELAXED) { + int r = rec; + while (--r >= 0) + if (strcmp(tok, root_names[r]) == 0) { + lprintf("warning: wanted %s, got %s - assuming missing close tags\n", root_names[rec], tok); + retval = r - rec - 1; /* -1 - (no. of implied close tags) */ + state = STATE_TAG_TERM; + break; + } + /* relaxed parsing, ignoring extra close tag (but we don't handle out-of-order) */ + if (r < 0) { + lprintf("warning: extra close tag %s - ignoring\n", tok); + state = STATE_TAG_TERM_IGNORE; + } + } + else + { + lprintf("error: xml struct, tok=%s, waited_tok=%s\n", tok, root_names[rec]); return -1; } break; @@ -324,10 +442,10 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r break; /* > expected */ - case 4: + case STATE_TAG_TERM: switch (res) { case (T_M_STOP_1): - return 0; + return retval; break; default: lprintf("error: unexpected token \"%s\", state %d\n", tok, state); @@ -337,18 +455,18 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r break; /* = or > or ident or separator expected */ - case 5: + case STATE_ATTRIBUTE_EQUALS: switch (res) { case (T_EOL): case (T_SEPAR): /* do nothing */ break; case (T_EQUAL): - state = 6; + state = STATE_STRING; break; case (T_IDENT): bypass_get_token = 1; /* jump to state 2 without get a new token */ - state = 2; + state = STATE_ATTRIBUTE; break; case (T_M_STOP_1): /* add a new property without value */ @@ -362,7 +480,42 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r current_property->name = strdup (property_name); lprintf("info: new property %s\n", current_property->name); bypass_get_token = 1; /* jump to state 2 without get a new token */ - state = 2; + state = STATE_ATTRIBUTE; + break; + default: + lprintf("error: unexpected token \"%s\", state %d\n", tok, state); + return -1; + break; + } + break; + + /* = or ?> or ident or separator expected */ + case STATE_Q_ATTRIBUTE_EQUALS: + switch (res) { + case (T_EOL): + case (T_SEPAR): + /* do nothing */ + break; + case (T_EQUAL): + state = STATE_Q_STRING; + break; + case (T_IDENT): + bypass_get_token = 1; /* jump to state 2 without get a new token */ + state = STATE_Q_ATTRIBUTE; + break; + case (T_TI_STOP): + /* add a new property without value */ + if (current_property == NULL) { + properties = new_xml_property(); + current_property = properties; + } else { + current_property->next = new_xml_property(); + current_property = current_property->next; + } + current_property->name = strdup (property_name); + lprintf("info: new property %s\n", current_property->name); + bypass_get_token = 1; /* jump to state 2 without get a new token */ + state = STATE_Q_ATTRIBUTE; break; default: lprintf("error: unexpected token \"%s\", state %d\n", tok, state); @@ -372,7 +525,8 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r break; /* string or ident or separator expected */ - case 6: + case STATE_STRING: + case STATE_Q_STRING: switch (res) { case (T_EOL): case (T_SEPAR): @@ -391,7 +545,7 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r current_property->name = strdup(property_name); current_property->value = lexer_decode_entities(tok); lprintf("info: new property %s=%s\n", current_property->name, current_property->value); - state = 2; + state = Q_STATE(STRING, ATTRIBUTE); break; default: lprintf("error: unexpected token \"%s\", state %d\n", tok, state); @@ -407,31 +561,45 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r state = STATE_IDLE; break; default: - state = STATE_COMMENT; break; } break; - /* ?> expected */ - case 8: + /* > expected */ + case STATE_DOCTYPE: switch (res) { - case (T_TI_STOP): + case (T_M_STOP_1): state = 0; break; default: - state = 8; break; } break; - /* > expected */ - case 9: + /* ]]> expected */ + case STATE_CDATA: + switch (res) { + case (T_CDATA_STOP): + current_subtree = xml_parser_append_text (current_node, current_subtree, tok, flags); + lprintf("info: node cdata : %s\n", tok); + state = STATE_IDLE; + break; + default: + lprintf("error: unexpected token \"%s\", state %d\n", tok, state); + return -1; + break; + } + break; + + /* > expected (following unmatched "</...") */ + case STATE_TAG_TERM_IGNORE: switch (res) { case (T_M_STOP_1): - state = 0; + state = STATE_IDLE; break; default: - state = 9; + lprintf("error: unexpected token \"%s\", state %d\n", tok, state); + return -1; break; } break; @@ -452,14 +620,51 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r } } -int xml_parser_build_tree(xml_node_t **root_node) { - xml_node_t *tmp_node; +static int xml_parser_get_node (xml_node_t *current_node, int flags) +{ + char *root_names[MAX_RECURSION + 1]; + root_names[0] = ""; + return xml_parser_get_node_internal (current_node, root_names, 0, flags); +} + +int xml_parser_build_tree_with_options(xml_node_t **root_node, int flags) { + xml_node_t *tmp_node, *pri_node, *q_node; int res; tmp_node = new_xml_node(); - res = xml_parser_get_node(tmp_node, "", 0); - if ((tmp_node->child) && (!tmp_node->child->next)) { - *root_node = tmp_node->child; + res = xml_parser_get_node(tmp_node, flags); + + /* delete any top-level [CDATA] nodes */; + pri_node = tmp_node->child; + q_node = NULL; + while (pri_node) { + if (pri_node->name == cdata) { + xml_node_t *old = pri_node; + if (q_node) + q_node->next = pri_node->next; + else + q_node = pri_node; + pri_node = pri_node->next; + free_xml_node (old); + } else { + q_node = pri_node; + pri_node = pri_node->next; + } + } + + /* find first non-<?...?> node */; + for (pri_node = tmp_node->child, q_node = NULL; + pri_node && pri_node->name[0] == '?'; + pri_node = pri_node->next) + q_node = pri_node; /* last <?...?> node (eventually), or NULL */ + + if (pri_node && !pri_node->next) { + /* move the tail to the head (for compatibility reasons) */ + if (q_node) { + pri_node->next = tmp_node->child; + q_node->next = NULL; + } + *root_node = pri_node; free_xml_node(tmp_node); res = 0; } else { @@ -470,6 +675,10 @@ int xml_parser_build_tree(xml_node_t **root_node) { return res; } +int xml_parser_build_tree(xml_node_t **root_node) { + return xml_parser_build_tree_with_options (root_node, 0); +} + const char *xml_parser_get_property (const xml_node_t *node, const char *name) { xml_property_t *prop; @@ -588,5 +797,8 @@ static void xml_parser_dump_node (const xml_node_t *node, int indent) { } void xml_parser_dump_tree (const xml_node_t *node) { - xml_parser_dump_node (node, 0); + do { + xml_parser_dump_node (node, 0); + node = node->next; + } while (node); } |