Merged in feature/checkbox (pull request #145)

Checkbox Primary Edge Cases

* hascheckboxlogic

* gen

* preppinggen

* exploringcheckbox

* removeunselected

* tests

* failingtest

* failintests

* nomockqueryapi

* db

* name

* nocontainer

* deleterow

* cnc

* extradocsandupdatetextracts

* moretables

* appndedtables

* baseversion
This commit is contained in:
Michael McGuinness
2025-05-20 12:47:22 +00:00
parent beb221937b
commit 61d12bf8df
69 changed files with 223611 additions and 78482 deletions
+97 -29
View File
@@ -66,6 +66,82 @@ func TestGetDocumentText(t *testing.T) {
_, err := svc.getDocumentText(ctx, pdf)
require.EqualError(t, err, "no textract response")
})
t.Run("merged_cell_table", func(t *testing.T) {
t.Parallel()
cfg := DocTextConfig{}
svc := Service{
cfg: &cfg,
}
mockTextract := textractmock.NewMockTextractClient(t)
cfg.TextractClient = mockTextract
textractBaseOut, pdf, txtFile, close := getFiles(t, "merged_cell_table")
defer close()
textExpectations(t, ctx, mockTextract, pdf, textractBaseOut)
text, err := svc.getDocumentText(ctx, pdf)
require.NoError(t, err)
assertReaders(t, txtFile, text)
})
t.Run("table_check", func(t *testing.T) {
t.Parallel()
cfg := DocTextConfig{}
svc := Service{
cfg: &cfg,
}
mockTextract := textractmock.NewMockTextractClient(t)
cfg.TextractClient = mockTextract
textractBaseOut, pdf, txtFile, close := getFiles(t, "table_check")
defer close()
textExpectations(t, ctx, mockTextract, pdf, textractBaseOut)
text, err := svc.getDocumentText(ctx, pdf)
require.NoError(t, err)
assertReaders(t, txtFile, text)
})
t.Run("multi_column", func(t *testing.T) {
t.Parallel()
cfg := DocTextConfig{}
svc := Service{
cfg: &cfg,
}
mockTextract := textractmock.NewMockTextractClient(t)
cfg.TextractClient = mockTextract
textractBaseOut, pdf, txtFile, close := getFiles(t, "multi_column")
defer close()
textExpectations(t, ctx, mockTextract, pdf, textractBaseOut)
text, err := svc.getDocumentText(ctx, pdf)
require.NoError(t, err)
assertReaders(t, txtFile, text)
})
t.Run("table_check_row", func(t *testing.T) {
t.Parallel()
cfg := DocTextConfig{}
svc := Service{
cfg: &cfg,
}
mockTextract := textractmock.NewMockTextractClient(t)
cfg.TextractClient = mockTextract
textractBaseOut, pdf, txtFile, close := getFiles(t, "table_check_row")
defer close()
textExpectations(t, ctx, mockTextract, pdf, textractBaseOut)
text, err := svc.getDocumentText(ctx, pdf)
require.NoError(t, err)
assertReaders(t, txtFile, text)
})
t.Run("cnc_01-06", func(t *testing.T) {
t.Parallel()
@@ -141,25 +217,6 @@ func TestGetDocumentText(t *testing.T) {
text, err := svc.getDocumentText(ctx, pdf)
require.NoError(t, err)
assertReaders(t, txtFile, text)
})
t.Run("merged_cell_table", func(t *testing.T) {
t.Parallel()
cfg := DocTextConfig{}
svc := Service{
cfg: &cfg,
}
mockTextract := textractmock.NewMockTextractClient(t)
cfg.TextractClient = mockTextract
textractBaseOut, pdf, txtFile, close := getFiles(t, "merged_cell_table")
defer close()
textExpectations(t, ctx, mockTextract, pdf, textractBaseOut)
text, err := svc.getDocumentText(ctx, pdf)
require.NoError(t, err)
assertReaders(t, txtFile, text)
})
}
@@ -208,13 +265,20 @@ func textExpectations(t testing.TB, ctx context.Context, mockTextract *textractm
).
Return(base, nil)
table := pageResult["table"]
table := pageResult["full_features"]
if table != nil {
mockTextract.EXPECT().
AnalyzeDocument(
mock.Anything,
mock.MatchedBy(func(in *textract.AnalyzeDocumentInput) bool {
return string(in.Document.Bytes) == string(page) && in.FeatureTypes[0] == types.FeatureTypeTables
found := false
for _, feature := range in.FeatureTypes {
if feature == types.FeatureTypeForms || feature == types.FeatureTypeTables {
found = true
break
}
}
return string(in.Document.Bytes) == string(page) && found
}),
mock.Anything,
).
@@ -223,17 +287,21 @@ func textExpectations(t testing.TB, ctx context.Context, mockTextract *textractm
}
}
func assertReaderAndString(t testing.TB, expected io.Reader, actual string) {
expectedBytes, err := io.ReadAll(expected)
require.NoError(t, err)
expectedStr := string(expectedBytes)
if !assert.Equal(t, expectedStr, actual) {
diff := cmp.Diff(expectedStr, actual)
t.Logf("Detailed diff (-expected +actual):\n%s", diff)
}
}
func assertReaders(t testing.TB, expected io.Reader, actual io.Reader) {
actualBytes, err := io.ReadAll(actual)
require.NoError(t, err)
actualStr := string(actualBytes)
expectedBytes, err := io.ReadAll(expected)
require.NoError(t, err)
expectedStr := string(expectedBytes)
if !assert.Equal(t, expectedStr, actualStr) {
diff := cmp.Diff(expectedStr, actualStr)
t.Logf("Detailed diff (-expected +actual):\n%s", diff)
}
assertReaderAndString(t, expected, actualStr)
}