From d937a4994b54e88ce8cf875e4b30b13112ecac6a Mon Sep 17 00:00:00 2001
From: Philipp Zabel <philipp.zabel@gmail.com>
Date: Fri, 13 Aug 2010 22:28:56 +0200
Subject: [PATCH] IMDb line parser: add parser for actors and actresses lists

---
 src/imdb/imdb-line-parser.vala |  133 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 133 insertions(+)

diff --git a/src/imdb/imdb-line-parser.vala b/src/imdb/imdb-line-parser.vala
index 7963fda..43b6d96 100644
--- a/src/imdb/imdb-line-parser.vala
+++ b/src/imdb/imdb-line-parser.vala
@@ -312,3 +312,136 @@ class PlotLineParser : LineParser {
 		}
 	}
 }
+
+class PersonParser : LineParser {
+	enum PersonState {
+		HEADER,
+		NONE,
+		PERSON
+	}
+	enum PersonType {
+		NONE,
+		ACTOR,
+		ACTRESS,
+		DIRECTOR,
+		WRITER
+	}
+	PersonState state;
+	PersonType type;
+	string name;
+
+	public PersonParser (IMDbSqlite _sqlite) {
+		base (_sqlite);
+		reset ();
+	}
+
+	public void reset () {
+		state = PersonState.HEADER;
+		type = PersonType.NONE;
+		name = null;
+	}
+
+	public override void parse_line (string line) {
+		if (state == PersonState.HEADER) {
+			if (line == "THE ACTORS LIST") type = PersonType.ACTOR;
+			else if (line == "THE ACTRESSES LIST") type = PersonType.ACTRESS;
+			else if (line == "THE DIRECTORS LIST") type = PersonType.DIRECTOR;
+			else if (line == "THE WRITERS LIST") type = PersonType.WRITER;
+			else if (line == "----\t\t\t------" && type != PersonType.NONE)
+				state = PersonState.NONE;
+
+			return;
+		}
+
+		// Skip empty lines
+		if (line == "") {
+			state = PersonState.NONE;
+			name = null;
+
+			return;
+		}
+
+		if (state == PersonState.NONE) {
+			if (line.has_prefix ("\t")) {
+				stderr.printf ("Invalid person entry: %s\n", line);
+				error ("EXIT\n");
+				return;
+			}
+
+			unowned string title = line.rstr ("\t");
+			if (title == null)
+				return;
+			title = title.offset (1);
+
+			char* end = line.str ("\t");
+			if (end != null)
+				end[0] = '\0';
+
+			try {
+				name = convert (line, -1, "utf-8", "latin1");
+			} catch (ConvertError e) {
+				stderr.printf ("Error converting name to UTF-8\n");
+				name = null;
+
+				return;
+			}
+
+			sqlite.add_person (name);
+			parse_title (convert (title, -1, "utf-8", "latin1"));
+
+			state = PersonState.PERSON;
+
+			return;
+		}
+
+		if (state == PersonState.PERSON) {
+			if (line.has_prefix ("\t\t\t"))
+				parse_title (line.offset (3));
+			else
+				stderr.printf ("\t???: %s\n", line);
+		}
+	}
+
+	private void parse_title (string title) {
+		// Skip series episodes
+		if (title[0] == '"')
+			return;
+
+		char* end = title.str ("  ");
+		unowned string next;
+		if (end != null) {
+			next = ((string) end).offset (2);
+			end[0] = '\0';
+		} else {
+			return;
+		}
+
+		if (skip_title (title))
+			return;
+
+
+		if (type == PersonType.ACTOR || type == PersonType.ACTRESS) {
+			string character = null;
+			int number = 0;
+			string info = null;
+			do {
+				unowned string current = next;
+				end = current.str ("  ");
+				if (end != null) {
+					next = ((string) end).offset (2);
+					end[0] = '\0';
+				}
+
+				if (current.has_prefix ("["))
+					character = current.substring (1, current.length - 2);
+				if (current.has_prefix ("<"))
+					number = current.offset (1).to_int ();
+				if (current.has_prefix ("("))
+					info = current;
+			} while (end != null);
+
+			sqlite.add_actor (name, title, info, character, number);
+		}
+	}
+}
+
-- 
1.7.9.5