lzy/sci-gui-agent-benchmark
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Refactor evaluator functions in JSON examples to use URL pattern matching. Update expected URL formats to regex patterns for better validation in chrome evaluation examples.

Browse Source
This commit is contained in:
Timothyxxx
2025-10-01 19:20:06 +00:00
parent 9be518435c
commit 1572068035
2 changed files with 16 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
evaluation_examples/examples/chrome/a728a36e-8bf1-4bb6-9a03-ef039a5233f0.json
View File

@@ -43,7 +43,7 @@
"chrome"
],
"evaluator": {
"func": "is_expected_active_tab",
"func": "is_expected_url_pattern_match",
"result": {
"type": "active_url_from_accessTree",
"goto_prefix": "https://www."
@@ -51,8 +51,9 @@
"expected": {
"type": "rule",
"rules": {
"type": "url",
"url": "https://www.dmv.virginia.gov/licenses-ids/license/applying/eligibility"
"expected": [
"^https://(www\\.)?dmv\\.virginia\\.gov/licenses-ids/license/applying/eligibility"
]
}
}
},

26
evaluation_examples/examples/chrome/b070486d-e161-459b-aa2b-ef442d973b92.json
View File

@@ -44,37 +44,35 @@
],
"evaluator": {
"func": [
"exact_match",
"exact_match"
"is_expected_url_pattern_match",
"is_expected_url_pattern_match"
],
"conj": "or",
"result": [
{
"type": "url_dashPart",
"goto_prefix": "https://www.",
"partIndex": -1,
"needDeleteId": false,
"returnType": "string"
"type": "active_url_from_accessTree",
"goto_prefix": "https://www."
},
{
"type": "url_dashPart",
"goto_prefix": "https://www.",
"partIndex": -1,
"needDeleteId": false,
"returnType": "string"
"type": "active_url_from_accessTree",
"goto_prefix": "https://www."
}
],
"expected": [
{
"type": "rule",
"rules": {
"expected": "tamiflu.html#side-effects"
"expected": [
"^https://(www\\.)?drugs\\.com/tamiflu\\.html#side-effects"
]
}
},
{
"type": "rule",
"rules": {
"expected": "tamiflu-side-effects.html"
"expected": [
"^https://(www\\.)?drugs\\.com/sfx/tamiflu-side-effects\\.html"
]
}
}
]