Support filtering timezone names by regex

.github/workflows/rust.yml

@@ -30,18 +30,15 @@ jobs:
         git submodule sync --recursive
         git -c "http.extraheader=$auth_header" -c protocol.version=2 \
             submodule update --init --force --recursive --depth=1
-
-    - name: clippy
-      if: ${{ matrix.run_lint }}
-      run: cargo clippy --color=always
-    - name: rustfmt
-      if: ${{ matrix.run_lint }}
-      run: cargo fmt -- --color=always --check
-
     - name: Run tests
       run: cargo test --color=always -- --color=always
+
     - name: Run tests serde
       run: cargo test --features serde --color=always -- --color=always
+
+    - name: Run regex tests
+      run: bin/test-regex-filtering.sh
+
     - name: Check with no default features
       run: cargo check --no-default-features --color=always
 
@@ -64,3 +61,36 @@ jobs:
       run: |
         cargo clean
         cargo build --features serde1
+
+  lint:
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Checkout submodules
+      shell: bash
+      run: |
+        auth_header="$(git config --local --get http.https://github.com/.extraheader)"
+        git submodule sync --recursive
+        git -c "http.extraheader=$auth_header" -c protocol.version=2 \
+            submodule update --init --force --recursive --depth=1
+
+    - name: clippy
+      run: cargo clippy --color=always
+
+    - name: rustfmt
+      run: cargo fmt -- --color=always --check
+
+    # chrono-tz-build
+
+    - name: clippy chrono-tz-build
+      working-directory: ./chrono-tz-build
+      run: cargo clippy --color=always
+
+    - name: clippy chrono-tz-build filter-by-regex
+      working-directory: ./chrono-tz-build
+      run: cargo clippy --features filter-by-regex --color=always
+
+    - name: rustfmt
+      working-directory: ./chrono-tz-build
+      run: cargo fmt -- --color=always --check

CHANGELOG.md

@@ -0,0 +1,11 @@
+Chrono-tz Changelog
+===================
+
+## 0.6.0
+
+* **tzdb** [breaking change] Update tzdb to 2020b, which removes the `US/Pacific-New` timezone.
+  https://github.com/eggert/tz/commit/284e877d7511d964249ecfd2e75a9cab85e2741a
+
+* **feature** Add support for filtering the set of timezones with a new `filter-by-regex` feature
+  which uses `CHRONO_TZ_TIMEZONE_FILTER` env var. It should be set to a regular expression of
+  timezones to include.

Cargo.toml

@@ -17,9 +17,10 @@ serde = { version = "1", optional = true, default-features = false }
 [features]
 default = ["std"]
 std = []
+filter-by-regex = ["chrono-tz-build/filter-by-regex"]
 
 [build-dependencies]
-parse-zoneinfo = { version = "0.3" }
+chrono-tz-build = { path = "./chrono-tz-build", version = "0.0.1" }
 
 [dev-dependencies]
 serde_test = "1"

README.md

@@ -139,6 +139,7 @@ assert_eq!(utc.to_string(), "2016-10-21 23:00:00 UTC");
 
 To use this library without depending on the Rust standard library, put this
 in your `Cargo.toml`:
+
 ```toml
 [dependencies]
 chrono = { version = "0.4", default-features = false }
@@ -160,6 +161,32 @@ lto = true
 Otherwise, the additional binary size added by this library may overflow
 available program space and trigger a linker error.
 
+## Limiting the Timezone Table to Zones of Interest
+
+`Chrono-tz` by default generates timezones for all entries in the [IANA database][]. If you are
+interested in only a few timezones you can use enable the `filter-by-regex` feature and set an
+environment variable to select them. The environment variable is called
+`CHRONO_TZ_TIMEZONE_FILTER` and is a regular expression. It should be specified in your top-level
+build:
+
+```toml
+[dependencies]
+chrono-tz = { version = "0.6", features = [ "filter-by-regex" ] }
+```
+
+```sh
+CHRONO_TZ_TIMEZONE_FILTER="(Europe/London|US/.*)" cargo build
+```
+
+This can significantly reduce the size of the generated database, depending on how many timezones
+you are interested in. Wikipedia has an [article listing the timezone names][wiki-list].
+
+The filtering applied is liberal; if you use a pattern such as "US/.*" then `chrono-tz` will
+include all the zones that are linked, such as "America/Denver", not just "US/Mountain".
+
+[IANA database]: http://www.iana.org/time-zones
+[wiki-list]: https://en.wikipedia.org/wiki/List_of_tz_database_time_zones
+
 ## Future Improvements
 
 - Handle leap seconds

bin/test-regex-filtering.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+
+set -euxo pipefail
+
+export RUST_BACKTRACE=1
+export CHRONO_TZ_TIMEZONE_FILTER='(Europe/London|GMT)'
+
+cd tests/check-regex-filtering
+
+cargo test --color=always -- --color=always