feat: add support for IRI-reference and absolute-IRI

Signed-off-by: Jan Kowalleck <jan.kowalleck@gmail.com>

Author: jkowalleck

README.md

@@ -71,18 +71,25 @@ if is_valid_syntax(term='iri', value='http://github.com'):
 
 if not is_valid_syntax(term='iri', value='bob'):
     print("✗ Invalid IRI syntax")
+
+if not is_valid_syntax(term='iri_reference', value='bob'):
+    print("✓ Valid IRI-reference syntax")
 ```
 
 ### Alternatively, use term-specific helpers to validate RFC 3987 syntax.
 
 ```python
 from rfc3987_syntax import is_valid_syntax_iri
+from rfc3987_syntax import is_valid_syntax_iri_reference
 
 if is_valid_syntax_iri('http://github.com'):
     print("✓ Valid IRI syntax")
 
 if not is_valid_syntax_iri('bob'):
     print("✗ Invalid IRI syntax")
+    
+if is_valid_syntax_iri_reference('bob'):
+    print("✓ Valid IRI-reference syntax")
 ```
 
 ### Get the Lark parse tree for a syntax validation (useful for additional semantic validation)
@@ -114,8 +121,12 @@ This grammar was derived from:
 | Rule/Component       | Source     | Notes |
 |----------------------|------------|-------|
 | `iri`                | RFC 3987   | Top-level IRI rule |
+| `iri_reference`      | RFC 3987   | Top-level IRI Reference rule |
+| `absolute_iri`       | RFC 3987   | Top-level Absolute IRI rule |
 | `scheme`             | RFC 3986   | Referenced by RFC 3987 §2.2 |
 | `ihier_part`         | RFC 3987   | IRI-specific hierarchy |
+| `irelative_ref`      | RFC 3987   | IRI-specific relative ref |
+| `irelative_part`     | RFC 3987   | IRI-specific relative part |
 | `iauthority`         | RFC 3986   | Standard URI authority |
 | `ipath_abempty`      | RFC 3986   | Path format variant |
 | `ipath_absolute`     | RFC 3986   | Absolute path |

src/rfc3987_syntax/syntax_helpers.py

@@ -8,7 +8,11 @@
 RFC3987_SYNTAX_GRAMMAR_PATH: Path = Path(__file__).parent / "syntax_rfc3987.lark"
 RFC3987_SYNTAX_TERMS: list[str] = [
     "iri",
+    "iri_reference",
+    "absolute_iri",
     "scheme",
+    "irelative_ref",
+    "irelative_part"
     "ihier_part",
     "iauthority",
     "iuserinfo",
@@ -43,7 +47,7 @@
 
 grammar: str = load_grammar(RFC3987_SYNTAX_GRAMMAR_PATH)
 
-syntax_parser = Lark(grammar, start=["iri"], parser=RFC3987_SYNTAX_PARSER_TYPE)
+syntax_parser = Lark(grammar, start=["iri", "iri_reference", "absolute_iri"], parser=RFC3987_SYNTAX_PARSER_TYPE)
 
 
 def parse(term: str, value: str) -> ParseTree:
@@ -73,6 +77,14 @@ def syntax_validator(text):
 
 is_valid_syntax_iri = make_syntax_validator("iri")
 
+is_valid_syntax_iri_reference = make_syntax_validator("iri_reference")
+
+is_valid_syntax_absolute_iri = make_syntax_validator("absolute_iri")
+
+is_valid_syntax_irelative_ref = make_syntax_validator("irelative_ref")
+
+is_valid_syntax_irelative_part = make_syntax_validator("irelative_part")
+
 is_valid_syntax_ihier_part = make_syntax_validator("ihier_part")
 
 is_valid_syntax_iauthority = make_syntax_validator("iauthority")

src/rfc3987_syntax/syntax_rfc3987.lark

@@ -1,13 +1,24 @@
 
 iri: scheme ":" ihier_part ("?" iquery)? ("#" ifragment)?
 
+iri_reference: iri | irelative_ref
+
+absolute_iri: scheme ":" ihier_part ("?" iquery)?
+
 scheme: alpha (alpha | digit | "+" | "-" | ".")*
 
 ihier_part: "//" iauthority ipath_abempty
           | ipath_absolute
           | ipath_rootless
           | ipath_empty
 
+irelative_ref: irelative_part ("?" iquery)? ("#" ifragment)?
+
+irelative_part: "//" iauthority ipath_abempty
+              | ipath_absolute
+              | ipath_noscheme
+              | ipath_empty
+
 iauthority: (iuserinfo "@")? ihost (":" port)?
 
 iuserinfo: (iunreserved | pct_encoded | sub_delims | ":")*

tests/valid_syntax.json

@@ -24,7 +24,23 @@
     "ihier_part": [],
     "iprivate": [],
     "iquery": [],
-    "absolute_iri": [],
+    "absolute_iri": [
+        {
+            "value": "ftp://пример.испытание.example.com/файл.txt",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "https://www.example.com/über",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "https://exämple.example.org/pageセクション1",
+            "expect_lark": true,
+            "reason": ""
+        }
+    ],
     "gen_delims": [],
     "ifragment": [],
     "isegment_nz_nc": [],
@@ -40,11 +56,57 @@
             "reason": "Regex may match due to naive prefix pattern, but triple slash `http:///` is invalid due to empty authority.",
             "conforms_with_rfc3987_semantics": false,
             "semantics_notes": "Can occur when ireg_name is empty. To prevent this, change the ireg_name production rule to: ireg_name: (iunreserved | pct_encoded | sub_delims)+"
+        },
+        {
+            "value": "mailto:用户@例子.公司@example.com",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "ftp://пример.испытание.example.com/файл.txt",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "https://www.example.com/über",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "https://exämple.example.org/page#セクション1",
+            "expect_lark": true,
+            "reason": ""
+        }
+    ],
+    "iri_reference": [
+        {
+            "value": "/documents/नियम.pdf",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "../../фото/лошадь.jpg",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "index.html#संपर्क",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "doc/guide.html#überblick",
+            "expect_lark": true,
+            "reason": ""
+        },
+        {
+            "value": "пример/тест#часть2",
+            "expect_lark": true,
+            "reason": ""
         }
     ],
-    "iri_reference": [],
     "reserved": [],
     "scheme": [],
     "pct_encoded": [],
     "ipv4address": []
-}
+}