On Fri, 28 Dec 2012, Austin Clements <[email protected]> wrote: > This parses the subset of Xapian's boolean term quoting rules that are > used by make_boolean_term. This is provided as a generic string > utility, but will be used shortly in notmuch restore to parse and > optimize for ID queries. > --- > util/string-util.c | 55 > ++++++++++++++++++++++++++++++++++++++++++++++++++++ > util/string-util.h | 11 +++++++++++ > 2 files changed, 66 insertions(+) > > diff --git a/util/string-util.c b/util/string-util.c > index e4bea21..83b4953 100644 > --- a/util/string-util.c > +++ b/util/string-util.c > @@ -96,3 +96,58 @@ make_boolean_term (void *ctx, const char *prefix, const > char *term, > > return 0; > } > + > +int > +parse_boolean_term (void *ctx, const char *str, > + char **prefix_out, char **term_out) > +{ > + *prefix_out = *term_out = NULL; > + > + /* Parse prefix */ > + const char *pos = strchr (str, ':'); > + if (! pos) > + goto FAIL; > + *prefix_out = talloc_strndup (ctx, str, pos - str); > + ++pos; > + > + /* Implement de-quoting compatible with make_boolean_term. */ > + if (*pos == '"') { > + char *out = talloc_array (ctx, char, strlen (pos)); > + int closed = 0; > + *term_out = out; > + /* Skip the opening quote, find the closing quote, and > + * un-double doubled internal quotes. */ > + for (++pos; *pos; ) { > + if (*pos == '"') { > + ++pos; > + if (*pos != '"') { > + /* Found the closing quote. */ > + closed = 1; > + break; > + } > + } > + *out++ = *pos++; > + } > + /* Did the term terminate without a closing quote or is there > + * trailing text after the closing quote? */ > + if (!closed || *pos) > + goto FAIL; > + *out = '\0'; > + } else { > + const char *start = pos; > + /* Check for text after the boolean term. */ > + while (*pos > ' ' && *pos != ')') > + ++pos; > + if (*pos) > + goto FAIL; > + /* No trailing text; dup the string so the caller can free > + * it. */ > + *term_out = talloc_strdup (ctx, start); > + } > + return 0;
This looks like it will fail if a line contains trailing white space (and that seemed to be the case from my experiment). Is that wanted? If yes we may want to make the error message clearer as I got an error of the form Warning: cannot parse query: id:[email protected] and the fact that line contained trailing spaces was not clear. > + > + FAIL: > + talloc_free (*prefix_out); > + talloc_free (*term_out); > + return 1; > +} > diff --git a/util/string-util.h b/util/string-util.h > index b8844a3..43d49d0 100644 > --- a/util/string-util.h > +++ b/util/string-util.h > @@ -33,4 +33,15 @@ char *strtok_len (char *s, const char *delim, size_t *len); > int make_boolean_term (void *talloc_ctx, const char *prefix, const char > *term, > char **buf, size_t *len); > > +/* Parse a boolean term query produced by make_boolean_term, returning > + * the prefix in *prefix_out and the term in *term_out. *prefix_out > + * and *term_out will be talloc'd with context ctx. > + * > + * Return: 0 on success, non-zero on parse error (including trailing > + * data in str). > + */ I think it would be worth detailing what requirements a boolean term here should meet to be parsed. In particular, since make_boolean_term is more restrictive than Xapian we should say whether this is like make_boolean_term, like Xapian or somewhere in between. Best wishes Mark > +int > +parse_boolean_term (void *ctx, const char *str, > + char **prefix_out, char **term_out); > + > #endif > -- > 1.7.10.4 _______________________________________________ notmuch mailing list [email protected] http://notmuchmail.org/mailman/listinfo/notmuch
