Extract episode details, year and flags from subtitle

7 files changed, 107 insertions, 5 deletions

diff --git a/p2pvr/cron/importSchedume.xml b/p2pvr/cron/importSchedume.xml
index 767b8d2..e59403a 100644
--- a/p2pvr/cron/importSchedume.xml
+++ b/p2pvr/cron/importSchedume.xml
@@ -28,6 +28,10 @@
 					<param name="startTime"><value source="parent" attribute="startTime" depth="1" /></param>
 					<param name="stopTime"><value source="parent" attribute="stopTime" depth="1" /></param>
 					<param name="eventID"><value source="parent" attribute="eventID" depth="1" /></param>
+					<param name="episode"><value source="parent" attribute="episode" depth="1" /></param>
+					<param name="episodes"><value source="parent" attribute="episodes" depth="1" /></param>
+					<param name="year"><value source="parent" attribute="year" depth="1" /></param>
+					<param name="flags"><value source="parent" attribute="flags" depth="1" /></param>
 				</parameters>
 			</p2:sqlmergeinsert>
 		</p2:iterate>
@@ -54,6 +58,10 @@
 			<column>contentSeriesID</column>
 			<column>startTime</column>
 			<column>stopTime</column>
+			<column>episode</column>
+			<column>episodes</column>
+			<column>year</column>
+			<column>flags</column>
 		</columns>
 	</p2:sqlmerge>
 </block>
diff --git a/p2pvr/cron/sched.xml b/p2pvr/cron/sched.xml
index 25e37d2..c2f9230 100644
--- a/p2pvr/cron/sched.xml
+++ b/p2pvr/cron/sched.xml
@@ -26,6 +26,8 @@
 			<column name="startTime" source="attribute">startTime</column>
 			<column name="stopTime" source="attribute">stopTime</column>
 			<column name="serviceID" source="attribute">serviceID</column>
+			<column name="episode" source="attribute">episode</column>
+			<column name="episodes" source="attribute">episodes</column>
 		</columns>
 	</p2:view>
 </block>
diff --git a/p2pvr/scanner/Jamfile.jam b/p2pvr/scanner/Jamfile.jam
index fca4615..81fdb9d 100644
--- a/p2pvr/scanner/Jamfile.jam
+++ b/p2pvr/scanner/Jamfile.jam
@@ -1,3 +1,5 @@
+lib boost_regex : : <name>boost_regex ;
+
 project
   : requirements
       <variant>debug:<linkflags>-Wl,-z,defs <cflags>"-W -Wall -Werror -Wwrite-strings"
@@ -7,7 +9,14 @@ project
 lib p2pvr-scan-ice : icescan.cpp scanner.ice : <library>p2pvr-scan-p2 ;
 
 # Scanner - t
-lib p2pvr-scan-p2 : eitRows.cpp serviceRows.cpp dvbSiReaderHelper.cpp : <library>../../project2//p2common ;
+lib p2pvr-scan-p2 :
+	eitRows.cpp
+	serviceRows.cpp
+	dvbSiReaderHelper.cpp
+	:
+	<library>../../project2//p2common
+	<library>boost_regex
+	;
 
 # ScannerICE - the ICE test app
 #exe scanice : ice_scan.cpp : <library>p2pvr-scan ;
diff --git a/p2pvr/scanner/dvbSiReaderHelper.cpp b/p2pvr/scanner/dvbSiReaderHelper.cpp
index a464520..330b1eb 100644
--- a/p2pvr/scanner/dvbSiReaderHelper.cpp
+++ b/p2pvr/scanner/dvbSiReaderHelper.cpp
@@ -46,7 +46,7 @@ DvbSiReaderHelper::closeInput() const
 	}
 }
 
-VariableType
+DvbSiReaderHelper::StrPtr
 DvbSiReaderHelper::convert(const char * txt, size_t len)
 {
 	if (len == 0) {
diff --git a/p2pvr/scanner/dvbSiReaderHelper.h b/p2pvr/scanner/dvbSiReaderHelper.h
index 1efb461..d82ab29 100644
--- a/p2pvr/scanner/dvbSiReaderHelper.h
+++ b/p2pvr/scanner/dvbSiReaderHelper.h
@@ -21,7 +21,8 @@ class DvbSiReaderHelper {
 		virtual bool parseInfoTable(const u_char *data, size_t len, const RowProcessor *) const = 0;
 		void closeInput() const;
 
-		static VariableType convert(const char * txt, size_t len);
+		typedef boost::shared_ptr<Glib::ustring> StrPtr;
+		static StrPtr convert(const char * txt, size_t len);
 		static const std::string ISO10646;
 		static const std::string EitEncoding;
 		static const std::string UTF8;
diff --git a/p2pvr/scanner/eitRows.cpp b/p2pvr/scanner/eitRows.cpp
index f5bcf51..23e8cfe 100644
--- a/p2pvr/scanner/eitRows.cpp
+++ b/p2pvr/scanner/eitRows.cpp
@@ -31,8 +31,14 @@
 #include <errno.h>
 #include <stdint.h>
 #include <time.h>
+#include <glibmm/regex.h>
+#include <boost/regex.hpp>
 #include <boost/bind.hpp>
 #include <boost/date_time/gregorian_calendar.hpp>
+#include <boost/algorithm/string/predicate.hpp>
+#include <boost/algorithm/string/trim.hpp>
+#include <boost/foreach.hpp>
+#include <boost/lexical_cast.hpp>
 
 #include <linux/dvb/dmx.h>
 #include "si_tables.h"
@@ -63,6 +69,10 @@ struct EitProgram {
 	VariableType contentRecommendation;
 	VariableType startTime;
 	VariableType stopTime;
+	VariableType episode;
+	VariableType episodes;
+	VariableType year;
+	VariableType flags;
 };
 
 static Glib::ustring title("title");
@@ -139,6 +149,10 @@ EitRows::resolveAttr(const Glib::ustring & attrName) const {
 	returnAttr(contentRecommendation);
 	returnAttr(startTime);
 	returnAttr(stopTime);
+	returnAttr(episode);
+	returnAttr(episodes);
+	returnAttr(year);
+	returnAttr(flags);
 	throw NoSuchAttribute(attrName);
 }
 
@@ -158,6 +172,9 @@ EitRows::~EitRows()
 static int time_offset   = 0;
 static int chan_filter	     = 0;
 static int chan_filter_mask = 0;
+static Glib::RefPtr<Glib::Regex> episodeRegex = Glib::Regex::create("[ (]+(?:\\w+ )?([0-9]+)(?: of |/)([0-9]+)[.)]+");
+static Glib::RefPtr<Glib::Regex> yearRegex = Glib::Regex::create("\\(([0-9]{4})[ )]+");
+static Glib::RefPtr<Glib::Regex> flagsRegex = Glib::Regex::create("[ []+([A-Z,]+)\\]");
 
 void
 EitRows::parseEventDescription(const u_char *data) const {
@@ -165,14 +182,77 @@ EitRows::parseEventDescription(const u_char *data) const {
 	const struct descr_short_event *evtdesc = reinterpret_cast<const struct descr_short_event *>(data);
 
 	size_t evtlen = evtdesc->event_name_length;
+	StrPtr title, subtitle, desc;
 	if (evtlen) {
 		current->titleLang = Glib::ustring((const char *)&evtdesc->lang_code1, 3);
-		current->title = convert((const char *)evtdesc->data, evtlen);
+		title = convert((const char *)evtdesc->data, evtlen);
 	}
 
 	size_t dsclen = evtdesc->data[evtlen];
 	if (dsclen) {
-		current->subtitle = convert((const char *)evtdesc->data + evtlen + 1, dsclen);
+		subtitle = convert((const char *)evtdesc->data + evtlen + 1, dsclen);
+	}
+	if (subtitle) {
+		Glib::MatchInfo matches;
+		if (episodeRegex->match(*subtitle, matches)) {
+			current->episode = boost::lexical_cast<int>(matches.fetch(1));
+			current->episodes = boost::lexical_cast<int>(matches.fetch(2));
+			*subtitle = episodeRegex->replace_literal(*subtitle, 0, "", Glib::REGEX_MATCH_NOTEMPTY);
+		}
+		if (yearRegex->match(*subtitle, matches)) {
+			current->year = boost::lexical_cast<int>(matches.fetch(1));
+			*subtitle = yearRegex->replace_literal(*subtitle, 0, "", Glib::REGEX_MATCH_NOTEMPTY);
+		}
+		if (flagsRegex->match(*subtitle, matches)) {
+			current->flags = matches.fetch(1);
+			*subtitle = yearRegex->replace_literal(*subtitle, 0, "", Glib::REGEX_MATCH_NOTEMPTY);
+		}
+	}
+	if (title && subtitle) {
+		if (boost::algorithm::ends_with(*title, "...") && boost::algorithm::starts_with(*subtitle, "...")) {
+			title->resize(title->length() - 3);
+			*title += " ";
+			size_t dot = subtitle->find('.', 4);
+			if (dot == Glib::ustring::npos) {
+				title->append(*subtitle, 3, subtitle->length() - 3);
+			}
+			else {
+				title->append(*subtitle, 3, dot - 3);
+				subtitle->erase(0, dot + 2);
+			}
+		}
+		size_t colon = subtitle->find(':');
+		if (colon != Glib::ustring::npos) {
+			desc = StrPtr(new Glib::ustring(*subtitle, colon + 1, subtitle->length() - colon));
+			subtitle->resize(colon);
+		}
+		else {
+			colon = title->find(':');
+			desc = subtitle;
+			if (colon != Glib::ustring::npos) {
+				subtitle = StrPtr(new Glib::ustring(*title, colon + 1, title->length() - colon));
+				title->resize(colon);
+			}
+			else {
+				subtitle.reset();
+			}
+		}
+	}
+	if (title) {
+		boost::algorithm::trim_if(*title, isspace);
+		current->title = title;
+	}
+	if (subtitle) {
+		boost::algorithm::trim_if(*subtitle, isspace);
+		if (!subtitle->empty()) {
+			current->subtitle = subtitle;
+		}
+	}
+	if (desc) {
+		boost::algorithm::trim_if(*desc, isspace);
+		if (!desc->empty()) {
+			current->desc1 = desc;
+		}
 	}
 }
 
diff --git a/p2pvr/scanner/si_tables.h b/p2pvr/scanner/si_tables.h
index 3721db4..c48fdae 100644
--- a/p2pvr/scanner/si_tables.h
+++ b/p2pvr/scanner/si_tables.h
@@ -28,6 +28,8 @@
 // Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 // Boston, MA 02111-1307, USA.
 
+#include <stdint.h>
+
 #define HILO(x) (x##_hi << 8 | x##_lo)
 #define HILO2(x) (x##1 << 8 | x##2)
 #define HILO3(x) (x##1 << 16 | x##2 << 8 | x##3)