X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=parse_trigrams.h;h=c845cdeab93cec0f0803b01d01cda232e422b546;hb=a2a3c6f0a7037c012839c5b24523b9f9e3f6f195;hp=810d005653fcf7ab3906f4ddc006670f3e99a4ec;hpb=efd7545c8ee2177aa13cf3ec8423d0e725c6a16d;p=plocate

diff --git a/parse_trigrams.h b/parse_trigrams.h
index 810d005..c845cde 100644
--- a/parse_trigrams.h
+++ b/parse_trigrams.h
@@ -11,7 +11,7 @@
 // One or more trigrams, with an implicit OR between them. For case-sensitive searches,
 // this is just e.g. “abc”, but for case-insensitive, it would be “abc OR abC or aBc ...” etc.
 struct TrigramDisjunction {
-	unsigned index;  // For debugging only.
+	unsigned index = -1;  // For debugging only.
 
 	// The alternatives as determined by parse_trigrams().
 	std::vector<uint32_t> trigram_alternatives;
@@ -53,6 +53,19 @@ struct TrigramDisjunction {
 // getting their own trigram).
 void parse_trigrams(const std::string &needle, bool ignore_case, std::vector<TrigramDisjunction> *trigram_groups);
 
+static constexpr uint32_t WILDCARD_UNIGRAM = 0xFF000000;
+static constexpr uint32_t PREMATURE_END_UNIGRAM = 0xFF000001;
+
+// Reads a unigram, taking into account escaping (\<foo> becomes <foo>).
+// Returns WILDCARD_UNIGRAM if there's an invalid unigram, ie., we found
+// a glob character (?, * or a [] group). Returns EOS_UNIGRAM if we went
+// past the end of the string, e.g., a string that ends in a backslash.
+// The second element is always the length.
+std::pair<uint32_t, size_t> read_unigram(const std::string &s, size_t start);
+
+// Reads a trigram, ie., three calls to read_unigram(). Needs to start on a valid unigram.
+// Returns WILDCARD_UNIGRAM or PREMATURE_END_UNIGRAM of either of those occurred
+// during reading of the string.
 uint32_t read_trigram(const std::string &s, size_t start);
 
 // For debugging.