Merge pull request #151 from kylerchin/main

Add ability to skip shapes reading

Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 description = "Read GTFS (public transit timetables) files"
 name = "gtfs-structures"
-version = "0.38.0"
+version = "0.39.0"
 authors = ["Tristram Gräbener <[email protected]>", "Antoine Desbordes <[email protected]>"]
 repository = "https://github.com/rust-transit/gtfs-structure"
 license = "MIT"

src/gtfs.rs

@@ -9,7 +9,7 @@ use std::sync::Arc;
 ///
 /// This structure is easier to use than the [RawGtfs] structure as some relationships are parsed to be easier to use.
 ///
-/// If you want to configure the behaviour (e.g. skipping : [StopTime]), see [crate::GtfsReader] for more personalisation
+/// If you want to configure the behaviour (e.g. skipping : [StopTime] or [Shape]), see [crate::GtfsReader] for more personalisation
 ///
 /// This is probably the entry point you want to use:
 /// ```

src/gtfs_reader.rs

@@ -14,6 +14,7 @@ use std::path::Path;
 /// ```
 ///let gtfs = gtfs_structures::GtfsReader::default()
 ///    .read_stop_times(false) // Won’t read the stop times to save time and memory
+///    .read_shapes(false) // Won’t read shapes to save time and memory
 ///    .unkown_enum_as_default(false) // Won’t convert unknown enumerations into default (e.g. LocationType=42 considered as a stop point)
 ///    .read("fixtures/zips/gtfs.zip")?;
 ///assert_eq!(0, gtfs.trips.get("trip1").unwrap().stop_times.len());
@@ -36,12 +37,15 @@ pub struct GtfsReader {
     /// [crate::objects::StopTime] are very large and not always needed. This allows to skip reading them
     #[derivative(Default(value = "true"))]
     pub read_stop_times: bool,
-    /// If a an enumeration has un unknown value, should we use the default value
+    /// [crate::objects::Shape] are very large and not always needed. This allows to skip reading them
+    #[derivative(Default(value = "true"))]
+    pub read_shapes: bool,
+    /// If a an enumeration has an unknown value, should we use the default value
     #[derivative(Default(value = "false"))]
     pub unkown_enum_as_default: bool,
     /// Avoid trimming the fields
     ///
-    /// It is quite time consumming
+    /// It is quite time consuming
     /// If performance is an issue, and if your data is high quality, you can switch it off
     #[derivative(Default(value = "true"))]
     pub trim_fields: bool,
@@ -57,6 +61,13 @@ impl GtfsReader {
         self
     }
 
+    /// This can be useful to save time and memory with large datasets when shapes are not needed
+    /// Returns Self and can be chained
+    pub fn read_shapes(mut self, read_shapes: bool) -> Self {
+        self.read_shapes = read_shapes;
+        self
+    }
+
     /// If a an enumeration has un unknown value, should we use the default value (default: false)
     ///
     /// For instance, if [crate::objects::Stop] has a [crate::objects::LocationType] with a value 42 in the GTFS
@@ -290,7 +301,11 @@ impl RawGtfsReader {
             transfers: self.read_optional_file(&file_mapping, &mut archive, "transfers.txt"),
             pathways: self.read_optional_file(&file_mapping, &mut archive, "pathways.txt"),
             feed_info: self.read_optional_file(&file_mapping, &mut archive, "feed_info.txt"),
-            shapes: self.read_optional_file(&file_mapping, &mut archive, "shapes.txt"),
+            shapes: if self.reader.read_shapes {
+                self.read_optional_file(&file_mapping, &mut archive, "shapes.txt")
+            } else {
+                Some(Ok(Vec::new()))
+            },
             read_duration: Utc::now().signed_duration_since(now).num_milliseconds(),
             files,
             source_format: crate::SourceFormat::Zip,