Merged in feature/postprocessing (pull request #114)
Feature/postprocessing * tests * passtest * fixshorttests * mosttests * improvingbasedockerfile * testspeeds * testing * host * canparallel * clean * passfullsuite * singlepagemax * test * findfeatures * findstables * tbls * tablestoo * tablestoo * lateraltests * tableloc * cleanup * inlinetable * childids * cleanup * tests
This commit is contained in:
@@ -9,17 +9,40 @@ import (
|
||||
"fmt"
|
||||
"log/slog"
|
||||
"os"
|
||||
documenttext "queryorchestration/internal/document/text"
|
||||
documenttypes "queryorchestration/internal/document/types"
|
||||
"queryorchestration/internal/serviceconfig"
|
||||
"queryorchestration/internal/serviceconfig/aws"
|
||||
"queryorchestration/internal/serviceconfig/objectstore"
|
||||
"queryorchestration/internal/serviceconfig/queue/querysync"
|
||||
"queryorchestration/internal/serviceconfig/textract"
|
||||
|
||||
awstextract "github.com/aws/aws-sdk-go-v2/service/textract"
|
||||
"github.com/aws/aws-sdk-go-v2/service/textract/types"
|
||||
)
|
||||
|
||||
type Config struct {
|
||||
serviceconfig.BaseConfig
|
||||
textract.TextractConfig
|
||||
querysync.QuerySyncConfig
|
||||
objectstore.ObjectStoreConfig
|
||||
}
|
||||
|
||||
func main() {
|
||||
ctx := context.Background()
|
||||
|
||||
cfg := &textract.TextractConfig{}
|
||||
if len(os.Args) != 3 {
|
||||
fmt.Println("Error: expected at least two arguments")
|
||||
fmt.Println("Usage:", os.Args[0], "<sample_name>", "<file_name>")
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
sampleName := os.Args[1]
|
||||
name := os.Args[2]
|
||||
|
||||
slog.Info("arguments", "sample_name", sampleName, "file_name", name)
|
||||
|
||||
cfg := &Config{}
|
||||
profile := aws.Profile(os.Getenv("AWS_PROFILE"))
|
||||
if profile == "" {
|
||||
slog.Error("profile not available")
|
||||
@@ -32,35 +55,105 @@ func main() {
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
base := "../../assets/sampleGeneration"
|
||||
name := "helloWorld"
|
||||
filename := fmt.Sprintf("%s/%s.pdf", base, name)
|
||||
content, err := os.ReadFile(filename)
|
||||
root, err := os.OpenRoot("assets")
|
||||
if err != nil {
|
||||
slog.Error("error writing file", "error", err)
|
||||
slog.Error("error opening root", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
slog.Info("detecting text", "filename", filename)
|
||||
|
||||
res, err := cfg.GetTextractClient().DetectDocumentText(ctx, &awstextract.DetectDocumentTextInput{
|
||||
Document: &types.Document{
|
||||
Bytes: content,
|
||||
},
|
||||
})
|
||||
root, err = root.OpenRoot(sampleName)
|
||||
if err != nil {
|
||||
slog.Error("error detecting text", "error", err)
|
||||
slog.Error("error opening sample", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
jsonContent, err := json.MarshalIndent(res, "", "\t")
|
||||
pdfFile, err := root.Open(fmt.Sprintf("%s.pdf", name))
|
||||
if err != nil {
|
||||
slog.Error("error opening pdf file", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
pdf, err := documenttypes.NewPDFFromReader(pdfFile)
|
||||
if err != nil {
|
||||
slog.Error("error creating pdf file", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
textSvc := documenttext.New(cfg)
|
||||
|
||||
count, err := pdf.GetPageCount(ctx)
|
||||
if err != nil {
|
||||
slog.Error("error getting page count", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
slog.Info("detecting text")
|
||||
|
||||
allResults := map[string][]awstextract.AnalyzeDocumentOutput{}
|
||||
allResults["base"] = make([]awstextract.AnalyzeDocumentOutput, count)
|
||||
allResults["table"] = []awstextract.AnalyzeDocumentOutput{}
|
||||
for i := range count {
|
||||
result, err := textSvc.GetBasePage(ctx, pdf, i)
|
||||
if err != nil {
|
||||
slog.Error("error detecting text", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
for _, block := range result.Blocks {
|
||||
feature := textSvc.GetBlockFeature(block)
|
||||
if feature == documenttext.FeatureTable {
|
||||
pageBytes, err := pdf.GetPageAsPNG(ctx, i)
|
||||
if err != nil {
|
||||
slog.Error("error getting page", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
res, err := cfg.GetTextractClient().AnalyzeDocument(ctx, &awstextract.AnalyzeDocumentInput{
|
||||
Document: &types.Document{
|
||||
Bytes: pageBytes,
|
||||
},
|
||||
FeatureTypes: []types.FeatureType{
|
||||
types.FeatureTypeTables,
|
||||
},
|
||||
})
|
||||
if err != nil {
|
||||
slog.Error("error detecting table", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
allResults["table"] = append(allResults["table"], *res)
|
||||
}
|
||||
}
|
||||
|
||||
allResults["base"][i] = result
|
||||
}
|
||||
|
||||
generatedBase := "generated"
|
||||
|
||||
if _, err := root.Stat(generatedBase); os.IsNotExist(err) {
|
||||
err := root.Mkdir(generatedBase, 0755)
|
||||
if err != nil {
|
||||
slog.Error("error creating directory", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
slog.Info("directory created", "directory", generatedBase)
|
||||
}
|
||||
|
||||
jsonContent, err := json.MarshalIndent(allResults, "", "\t")
|
||||
if err != nil {
|
||||
slog.Error("error marshaling json", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
|
||||
outputFilename := fmt.Sprintf("%s/generated/%s.gen", base, name)
|
||||
err = os.WriteFile(outputFilename, jsonContent, 0600)
|
||||
outputFilename := fmt.Sprintf("%s/%s.gen", generatedBase, name)
|
||||
file, err := root.Create(outputFilename)
|
||||
if err != nil {
|
||||
slog.Error("error creating file", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
defer file.Close()
|
||||
|
||||
_, err = file.Write(jsonContent)
|
||||
if err != nil {
|
||||
slog.Error("error writing file", "error", err)
|
||||
os.Exit(1)
|
||||
|
||||
Reference in New Issue
Block a user