lsp: Auto update examples index

Signed-off-by: Charlie Egan <charlie@styra.com>

.github/workflows/update-example-index.yaml

@@ -0,0 +1,67 @@
+# this workflow is used to update the internal/lsp/examples/index.json
+# file containing an index of the content available on
+# http://docs.styra.com/opa/rego-by-example
+name: Update Examples Index
+
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: '0 1 * * *' # Run daily at 1 AM UTC
+
+jobs:
+  update-examples-index:
+    env:
+      RQ_VERSION: v0.0.9
+    name: Update Examples Index
+    runs-on: ubuntu-22.04
+    permissions:
+      contents: write # this workflow updates repo contents
+    steps:
+      - name: Check out code
+        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+
+      - uses: open-policy-agent/setup-opa@34a30e8a924d1b03ce2cf7abe97250bbb1f332b5 # v2.2.0
+        with:
+          version: edge
+          static: true
+
+      - name: Set up go for rq
+        uses: actions/setup-go@0a12ed9d6a96ab950c8f026ed9f722fe0da7ef32 # v5.0.2
+        with:
+          go-version-file: go.mod
+
+      - name: Restore rq cache
+        id: cache-rq
+        uses: actions/cache@0c45773b623bea8c8e75f6c82b208c3cf94ea4f9 # v4.0.2
+        with:
+          path: ~/go/bin/rq
+          key: ${{ runner.os }}-${{ runner.arch }}-go-rq-${{ env.RQ_VERSION }}
+      - run: go install git.sr.ht/~charles/rq/cmd/rq@${{ env.RQ_VERSION }}
+        if: steps.cache-rq.outputs.cache-hit != 'true'
+      - name: Cache rq binary
+        if: steps.cache-rq.outputs.cache-hit != 'true'
+        uses: actions/cache@0c45773b623bea8c8e75f6c82b208c3cf94ea4f9 # v4.0.2
+        with:
+          path: ~/go/bin/rq
+          key: ${{ runner.os }}-${{ runner.arch }}-go-rq-${{ env.RQ_VERSION }}
+
+      - name: Fetch sitemap and convert to JSON
+        run: |
+          set -o pipefail
+
+          TEMP_DIR=$(mktemp -d)
+
+          curl -L https://docs.styra.com/sitemap.xml -o "$TEMP_DIR/sitemap.xml"
+
+          cat "$TEMP_DIR/sitemap.xml" | \
+            rq -i xml --indent "  " | \
+            opa eval 'data.process.symbols' \
+              -d build/workflows/update-example-index/process.rego \
+              --format=pretty \
+              --stdin-input | \
+            tee internal/lsp/examples/index.json
+
+      - name: Commit and push changes
+        uses: stefanzweifel/git-auto-commit-action@8621497c8c39c72f3e2a999a26b4ca1b5058a842 # v5.0.1
+        with:
+          commit_message: Update Imported Docs

build/workflows/update-example-index/process.rego

@@ -0,0 +1,62 @@
+package process
+
+import rego.v1
+
+symbols := {"keywords": _keywords, "builtins": _builtins}
+
+_keywords[name] := path if {
+	some p in _pages
+	p[0] == "keywords"
+
+	name := p[1]
+
+	path := concat("/", p)
+}
+
+_builtins[name] := path if {
+	some p in _pages
+	p[0] == "builtins"
+
+	l := count(p)
+
+	l == 2
+
+	name := p[1]
+
+	count({p |
+		some p in _pages
+		p[0] == "builtins"
+		p[1] == name
+	}) < 2
+
+	path := concat("/", p)
+}
+
+_builtins[name] := path if {
+	some p in _pages
+	p[0] == "builtins"
+
+	l := count(p)
+
+	l > 2
+
+	name := concat(
+		".",
+		[
+			replace(p[1], "_", "."),
+			concat(".", array.slice(p, 2, l)),
+		],
+	)
+
+	path := concat("/", p)
+}
+
+_prefix := "https://docs.styra.com/opa/rego-by-example/"
+
+_pages contains page if {
+	some url in input.urlset.url
+
+	startswith(url.loc, _prefix)
+
+	page := split(trim_prefix(url.loc, _prefix), "/")
+}

internal/lsp/examples/index.json

@@ -1,17 +1,17 @@
 {
-    "builtins": {
-        "contains": "builtins/contains",
-        "io.jwt.decode_verify": "builtins/io_jwt/decode_verify",
-        "regex.globs_match": "builtins/regex/globs_match",
-        "regex.match": "builtins/regex/match",
-        "regex.template_match": "builtins/regex/template_match",
-        "time.clock": "builtins/time/clock",
-        "time.format": "builtins/time/format",
-        "time.now_ns": "builtins/time/now_ns",
-        "time.parse_ns": "builtins/time/parse_ns"
-    },
-    "keywords": {
-        "contains": "keywords/contains",
-        "some": "keywords/some"
-    }
+  "builtins": {
+    "contains": "builtins/contains",
+    "io.jwt.decode_verify": "builtins/io_jwt/decode_verify",
+    "regex.globs_match": "builtins/regex/globs_match",
+    "regex.match": "builtins/regex/match",
+    "regex.template_match": "builtins/regex/template_match",
+    "time.clock": "builtins/time/clock",
+    "time.format": "builtins/time/format",
+    "time.now_ns": "builtins/time/now_ns",
+    "time.parse_ns": "builtins/time/parse_ns"
+  },
+  "keywords": {
+    "contains": "keywords/contains",
+    "some": "keywords/some"
+  }
 }