Merge pull request #24 from alix-tz/limb

Limb scenario

README.md

@@ -1,4 +1,4 @@
-[![MIT License](https://img.shields.io/static/v1?style=plastic&label=license&message=MIT&color=brightgreen)](LICENSE) [![Version](https://img.shields.io/static/v1?style=plastic&label=version&message=0.3.2&color=blue)]()
+[![MIT License](https://img.shields.io/static/v1?style=plastic&label=license&message=MIT&color=brightgreen)](LICENSE) [![Version](https://img.shields.io/static/v1?style=plastic&label=version&message=0.4.2&color=blue)]()
 
 # ASPYRE GT
 
@@ -37,9 +37,9 @@ Process essential information to run Aspyre
     [opt] :param vpadding: value to add to VPOS attr. in String nodes (int)
 ```
 
-> supported values for `scenario`: "tkb", "pdfalto"  
+> supported values for `scenario`: "tkb", "pdfalto", "limb"  
 
-> `vpadding` is only used in PDFALTO scenario
+> `vpadding` is only used in PDFALTO and LIMB scenarios
 
 ##### Transkribus to eScriptorium scenario with `aspyre.TkbToEs()`
 :warning: really not the best way to [transfer data between these two softwares](https://lectaurep.hypotheses.org/documentation/de-transkribus-a-escriptorium).

aspyre/aspyrelib/aspyre.py

@@ -13,9 +13,9 @@
 from tqdm import tqdm
 
 from .utils import utils
-from .manage import (manage_tkbtoes, manage_pdfaltotoes, zip)
+from .manage import (manage_tkbtoes, manage_pdfaltotoes, manage_limbtoes, zip)
 
-SUPPORTED_SCENARIOS = ["tkb", "pdfalto"]  # + ["limb", "finereader"]
+SUPPORTED_SCENARIOS = ["tkb", "pdfalto", "limb"]  # + ["finereader"]
 ARCHIVE_EXTENSIONS = ["zip"]
 
 
@@ -94,7 +94,7 @@ def __init__(self, scenario=None, source=None, destination=None, talkative=False
             if self.proceed():
                 # parsing vpadding
                 # only valid with PDFALTO scenario
-                if self.scenario != 'pdfalto':
+                if not self.scenario in ['pdfalto', 'limb']:
                     self.vpadding = 0
                 else:
                     self.vpadding = vpadding
@@ -105,10 +105,10 @@ def __init__(self, scenario=None, source=None, destination=None, talkative=False
                     self.padding = True
 
                 if self.talkative:
-                    if self.padding and self.scenario == 'pdfalto':
+                    if self.padding and self.scenario in ['pdfalto', 'limb']:
                         utils.report(f'Will add padding to y-axis coords in string nodes: {self.vpadding}\n---',
                                      "H")
-                    elif self.scenario == 'pdflato' and not self.padding:
+                    elif self.scenario in ['pdflato', 'limb'] and not self.padding:
                         utils.report(f"No modification made to y-axis coords in string nodes\n---", "H")
 
 
@@ -303,3 +303,84 @@ def __init__(self, args):
             self.args = None
             utils.report("Failed to run PdfaltoToEs: args must be an AspyreArgs object!\n===[!]===", "E")
 
+
+class LimbToEs():
+    def __init__(self, args):
+        """Handle a Limb to eScriptorium transformation scenario
+
+        :param args: essential information to run transformation scenario
+        :type args: AspyreArgs object
+        """
+        if isinstance(args, type(AspyreArgs(test_type=True))):
+            self.args = args
+            self.args.add_log("Starting LIMB transformation scenario.")
+
+            # 1. handling zip
+            self.unzipped_source = None
+            if self.args.source.split(".")[-1] in ARCHIVE_EXTENSIONS:
+                if self.args.talkative:
+                    utils.report("Source is an archive, running unzipping scenario.\n---", "H")
+                self.unzipped_source = zip.unzip_scenario(self.args.source, self.args.scenario)
+                if self.unzipped_source is False:
+                    self.args.execution_status = "Failed"
+                    self.add_log("Something went wrong while unpacking the source.")
+                    utils.report("Failing at unpacking the archive, Apsyre can't proceed.\n---", "E")
+                else:
+                    self.args.add_log("Successfully unzipped source.\n---")
+            else:
+                if self.args.talkative:
+                    utils.report("Source is not an archive.\n---", "H")
+
+            if self.args.proceed():
+                # 2. collecting data
+                package = utils.list_directory(self.unzipped_source)
+                self.alto_files, self.image_files = manage_limbtoes.locate_alto_and_image_files(package)
+
+                if self.alto_files is False:
+                    self.args.add_log("Couldn't find any XML file or any image file.")
+                    utils.report("Aspyre can't run without either of these.\n---", "E")
+                    self.args.execution_status = "Failed"
+                else:
+                    self.args.add_log("Successfully collected data.")
+
+            if self.args.proceed():
+                # 3. transforming files
+                if self.args.talkative:
+                    iterator = tqdm(self.alto_files, desc="Processing ALTO XML files", unit=' file')
+                else:
+                    iterator = self.alto_files
+                for file in iterator:
+                    processed = 0
+                    try:
+                        manage_limbtoes.handle_a_file(file, self)
+                    except Exception as e:
+                        if self.args.talkative:
+                            utils.report(f"Error while processing {file} :", "E")
+                            print(e)
+                        self.args.add_log(f"Failed to process {file}.")
+                    else:
+                        processed += 1
+                if processed == 0:
+                    self.args.execution_status = "Failed"
+                elif processed < len(self.alto_files):
+                    self.args.add_log(f"Successfully transformed {processed} out of {len(self.alto_files)}")
+                else:
+                    self.args.add_log(f"Successfully processed sources files!")
+
+            if self.args.proceed():
+                # 4. serve a zip file
+                try:
+                    zip.zip_dir(self.args.destination, self.unzipped_source)
+                except Exception as e:
+                    if self.args.talkative:
+                        print(e)
+                    self.args.execution_status = "Failed"
+                    self.args.add_log('Failed to zip output.')
+                else:
+                    utils.report("Task completed ✓", "S")
+                    self.args.execution_status = 'Finished'
+                    self.args.add_log('Aspyre ran Limb scenario successufully!')
+
+        else:
+            self.args = None
+            utils.report("Failed to run LimbToEs: args must be an AspyreArgs object!\n===[!]===", "E")