rfc: Implement generators for XML text and attribute

Author: tienvx

rfc/0000-implement-generators-for-xml-text-and-attribute.md

@@ -0,0 +1,175 @@
+---
+name: implement-generators-for-xml-text-and-attribute
+started: 2024-02-10
+pr: pact-foundation/roadmap#0000
+tracking_issue: pact-foundation/roadmap#0000
+---
+## Summary
+
+This RFC proposes implementing generators for bodies of type `application/xml`.
+
+## Motivation
+
+Currently if we define generators for text or attribute of XML's body, mock server will return this error:
+
+> Generators are not supported with XML
+
+And there will be an error message in the log:
+
+> UNIMPLEMENTED: Generators are not currently supported with XML
+
+To fix this error, we need to implement generators for XML.
+
+## Guide-level explanation
+
+### Text
+
+To define generator for XML element's text, we can use this JSON format:
+
+```json
+{
+  "name": "release-date",
+  "children": [
+    {
+      "pact:generator:type": "Date",
+      "matcher": {
+        "format": "dd-MM-yyyy",
+        "pact:matcher:type": "date"
+      },
+      "content": ""
+    }
+  ],
+  "attributes": []
+}
+```
+
+Mock server should return XML response like this:
+
+```xml
+<?xml version='1.0'?><movie><release-date>27-11-2012</release-date></movie>
+```
+
+### Attribute
+
+To define generator for XML element's attribute, we can use this JSON format:
+
+
+```json
+{
+  "name": "specs",
+  "children": [],
+  "attributes": {
+    "runtime": {
+      "pact:generator:type": "Regex",
+      "regex": "\\d+ hours [0-5]?[0-9] minutes",
+      "pact:matcher:type": "regex",
+      "value": ""
+    },
+    "aspect-ratio": {
+      "pact:generator:type": "Regex",
+      "regex": "([0-9]+[.])?[0-9]+:[0-9]+",
+      "pact:matcher:type": "regex",
+      "value": ""
+    },
+    "color": {
+      "pact:generator:type": "Regex",
+      "regex": "color|black&white",
+      "pact:matcher:type": "regex",
+      "value": ""
+    }
+  }
+}
+```
+
+Mock server should return XML response like this:
+
+```xml
+<?xml version='1.0'?><movie><specs runtime='1 hours 23 minutes' aspect-ratio='3:4' color='color'/></movie>
+```
+
+## Reference-level explanation
+
+### Implementation
+
+- Begin at root element
+    - Apply generator for each attribute
+    - Apply generator for each child element (recursively)
+    - Apply generator for each child text
+    - If there are no text & there is generator for text, then generate text and append to the element
+
+### Corner cases
+
+#### Text
+
+- Element has no text
+    - Generator: `RandomInt(0, 999)`
+    - Path: `$.a['#text']`
+    - Before: `<?xml version='1.0'?><a></a>`
+    - After: `<?xml version='1.0'?><a>123</a>`
+    - Note: New child text is append to the element
+- Element has a child element
+    - Generator: `RandomInt(0, 999)`
+    - Path: `$.a['#text']`
+    - Before: `<?xml version='1.0'?><a><b/></a>`
+    - After: `<?xml version='1.0'?><a><b/>123</a>`
+    - Note: New child text is append to the element
+- Element has mixed children
+    - Generator: `RandomInt(0, 999)`
+    - Path: `$.a['#text']`
+    - Before: `<?xml version='1.0'?><a><b/>-1</a>`
+    - After: `<?xml version='1.0'?><a><b/>123</a>`
+    - Note: New child text is append to the element
+- Element has multiple texts
+    - Generator: `RandomInt(0, 999)`
+    - Path: `$.a['#text']`
+    - Before: `<?xml version='1.0'?><a>-1<b/>-2</a>`
+    - After: `<?xml version='1.0'?><a>123<b/>234</a>`
+    - Note: We can't define index for the text (e.g. can't do this `$.a['#text'][0]` or this `$.a['#text'][1]`), the same generator is applied to every texts
+- Element has namespace
+    - Generator: `RandomInt(0, 999)`
+    - Path: `$.n:a['#text']`
+    - Before: `<?xml version='1.0'?><n:a xmlns:n='http://example.com/namespace'>-1</n:a>`
+    - After: `<?xml version='1.0'?><n:a xmlns:n='http://example.com/namespace'>123</n:a>`
+- Element has mixed namespace
+    - Generator: `RandomInt(0, 999)`
+    - Path: `$.root.n:a['#text']`
+    - Before: `<?xml version='1.0'?><root><n:a xmlns:n='http://example.com/namespace'>-1</n:a><a>-2</a></root>`
+    - After: `<?xml version='1.0'?><root><n:a xmlns:n='http://example.com/namespace'>123</n:a><a>-2</a></root>`
+
+
+#### Attribute
+
+- Attribute has namespace
+    - Generator: `RandomInt(0, 999)`
+    - Path: `$.a['@n:attr']`
+    - Before: `<?xml version='1.0'?><a n:attr='-1' xmlns:n='http://example.com/namespace'/>`
+    - After: `<?xml version='1.0'?><a n:attr='123' xmlns:n='http://example.com/namespace'/>`
+- Attribute has mixed namespace
+    - Generator: `RandomInt(0, 999)`
+    - Path: `$.a['@n:attr']`
+    - Before: `<?xml version='1.0'?><a n:attr='-1' attr='-2' xmlns:n='http://example.com/namespace'/>`
+    - After: `<?xml version='1.0'?><a n:attr='123' attr='-2' xmlns:n='http://example.com/namespace'/>`
+
+## Drawbacks
+
+None
+
+## Rationale and alternatives
+
+None
+
+## Unresolved questions
+
+None
+
+## Future possibilities
+
+In the future, when we can define index for the text, then we can define different matcher/generator for each text, then we can apply different generator for each text:
+
+- Generators:
+    - Path: `$.a['#text'][0]`
+      Generator: `RandomInt(0, 999)`
+    - Path: `$.a['#text'][1]`
+      Generator: `RandomString(3)`
+- Before: `<?xml version='1.0'?><a>-1<b/>-2</a>`
+- After: `<?xml version='1.0'?><a>123<b/>abc</a>`