public void CanExtractContentOrderText()
{
using (var document = PdfDocument.Open(GetFilename()))
{
foreach (var page in document.GetPages())
{
var text = ContentOrderTextExtractor.GetText(page);
Assert.NotNull(text);
}
}
}
public static void Run(string filePath)
{
using (var document = PdfDocument.Open(filePath))
{
foreach (var page in document.GetPages())
{
var text = ContentOrderTextExtractor.GetText(page, true);
Console.WriteLine(text);
}
}
}
public Document Parse(string name, Stream readStream)
{
// PdfDocument pdf = PdfDocument.Open(readStream);
List <string> textList = new List <string>();
using (var pdf = PdfDocument.Open(readStream)) {
foreach (var page in pdf.GetPages())
{
var lines = ContentOrderTextExtractor.GetText(page, true)
.Split(GetLineSeparator()).Select(line => line.Trim())
.Where(line => !line.Equals(""));
textList.AddRange(lines);
}
}
string[] texts = textList.ToArray();
Document doc = new Document();
// Positional lookup for "relativeTo" rule
Dictionary <string, int> extractedPosition = new Dictionary <string, int>();
// Get configs
JObject lookup = GetConfig()["lookup"] as JObject;
if (lookup is null)
{
throw new Exception("Invalid parser config: lookup is null");
}
// Use a queue for iterating through properties, since some text extraction could depend on others
Queue <PropertyInfo> propQueue = new Queue <PropertyInfo>(Document.Properties);
while (propQueue.Count > 0)
{
PropertyInfo docProp = propQueue.Dequeue();
JToken config = lookup[docProp.Name];
if (config is null)
{
continue;
}
JTokenType configType = config.Type;
// Rules based on string matching
if (configType.Equals(JTokenType.Object))
{
JObject rules = config as JObject;
string target = null;
int targetPos = -1;
string value = "";
if (rules["index"] != null)
{
target = texts[(int)rules["index"]];
}
else if (rules["startsWith"] != null)
{
JToken startsWith = rules["startsWith"];
if (startsWith.Type == JTokenType.String)
{
(target, targetPos) = findLine(texts, (text => text.StartsWith((string)rules["startsWith"])));
}
else
{
string[] startsWithes = startsWith.ToObject <string[]>();
(target, targetPos) = findLine(texts, text => startsWithes.Any(query => text.StartsWith(query)));
}
}
else if (rules["indexOf"] != null)
{
targetPos = Array.IndexOf(texts, (string)rules["indexOf"]);
if (targetPos < 0)
{
continue;
}
target = texts[targetPos];
}
else if (rules["relativeTo"] != null)
{
targetPos = extractedPosition.GetValueOrDefault((string)rules["relativeTo"], -1);
// Relative Member not parsed yet, requeue for processing
if (targetPos == -1)
{
propQueue.Enqueue(docProp);
continue;
}
target = texts[targetPos];
}
// Allow user to find specific line relative to somone common string like "Total Due"
if (rules["offset"] != null)
{
targetPos += (int)rules["offset"];
target = texts[targetPos];
}
if (target is null)
{
continue;
}
extractedPosition.Add(docProp.Name, targetPos);
// Further extraction with regex if needed, select specific group of text
if (rules["regex"] != null)
{
string regexPattern = (string)rules["regex"];
Regex regex = new Regex(regexPattern);
Match matched = regex.Match(target);
int group = (int?)rules["group"] ?? 0;
value = matched.Groups[group].Value;
}
else
{
value = target;
}
docProp.SetValue(doc, serializeValue(value, docProp.PropertyType));
}
}
return(doc);
}