ファイルに分けてみる

analyzeSegment.py

@@ -0,0 +1,53 @@
+from markers import Markers
+from segments.SOI import SOI
+from segments.EOI import EOI
+from segments.SOF import SOF
+from segments.DQT import DQT
+from segments.DHT import DHT
+from segments.COM import COM
+from segments.APP import APP
+from segments.SOS import SOS
+
+class segment(object):
+    def __init__(self):
+        super(segment, self).__init__()
+
+    def analyze(self, marker, body):
+        if marker == Markers.JUST_FF:
+            return None
+        if marker == Markers.SOI:
+            # SOI (FFD8) has no body
+            return SOI().analyze(marker, body)
+        elif marker == Markers.EOI:
+            # EOI (FFD9) has no body
+            return EOI().analyze(marker, body)
+        if marker in Markers.SOF:
+            # do sof
+            return SOF().analyze(marker, body)
+        elif marker == Markers.DQT:
+            # do dqt
+            return DQT().analyze(marker, body)
+        elif marker == Markers.DHT:
+            # do dht
+            return DHT().analyze(marker, body)
+        elif marker == Markers.DRI:
+            # do dri
+            pass
+        elif marker == Markers.COM:
+            # do com
+            return COM().analyze(marker, body)
+        elif marker in Markers.APP:
+            # do app
+            return APP().analyze(marker, body)
+        elif marker == Markers.SOS:
+            # do sos
+            return SOS().analyze(marker, body)
+        elif marker in Markers.RST:
+            # do rst
+            pass
+        else:
+            raise ValueError(str(b'\xFF') + str(marker.to_bytes(1, 'big')) + " is not supported marker")
+        return {
+            "segmentName": "Not Implemented",
+            "length": 0,
+        }

file.py

@@ -0,0 +1,52 @@
+from markers import Markers
+
+class File(object):
+    IS_SCANNING = False
+    IMAGES = []
+    IMG_DATA = []
+
+    def __init__(self, path):
+        super(File, self).__init__()
+        self.f = open(path, 'rb')
+
+    def __del__(self):
+        self.f.close()
+
+    def getData(self, count):
+        bt = list(self.f.read(count))
+        if self.IS_SCANNING:
+            self.IMG_DATA.extend(bt)
+        return bt
+
+    def __getSegmentBodyLength(self, length_for_length_data):
+        length = 0
+        for i in range(length_for_length_data):
+            length = length + (self.getData(1)[0] * (0x100 ** (length_for_length_data - i - 1)))
+        return length - length_for_length_data
+
+    def __getSegmentBody(self, length_for_length_data = 2):
+        body_length = self.__getSegmentBodyLength(length_for_length_data)
+        return self.getData(body_length)
+
+    # return dict
+    def getSegment(self, marker):
+        if marker not in Markers.REGISTERED_MARKERS:
+            raise Exception("unexpected marker found. it's not supported.")
+        if marker != Markers.JUST_FF:
+            # データの最後にマーカー（2ビット）がついているので末尾2ビット削除
+            self.IMG_DATA = self.IMG_DATA[0:len(self.IMG_DATA - 2)]
+            self.IMAGES.append(self.IMG_DATA)
+            self.IMG_DATA = []
+            self.IS_SCANNING = False
+        if marker in [Markers.JUST_FF, Markers.SOI, Markers.EOI]:
+            return {
+                "marker": marker,
+                "body": None,
+            }
+        body = self.__getSegmentBody()
+        if marker == Markers.SOS:
+            self.IS_SCANNING = True
+        return {
+                "marker": marker,
+                "body": body,
+            }

markers.py

@@ -0,0 +1,70 @@
+class Markers(object):
+    SOI = 0xd8
+    EOI = 0xd9
+    JUST_FF = 0x00
+
+    # cf. https://hp.vector.co.jp/authors/VA032610/JPEGFormat/marker/SOF.htm
+    SOF = [
+        0xC0, # ハフマン / ベースライン / 非差分
+        0xC1, # ハフマン / シーケンシャル / 非差分
+        0xC2, # ハフマン / プログレッシブ / 非差分
+        0xC3, # ハフマン / ロスレス / 非差分
+        0xC5, # ハフマン / シーケンシャル / 差分
+        0xC6, # ハフマン / プログレッシブ / 差分
+        0xC7, # ハフマン / ロスレス / 差分
+        0xC9, # 算術 / シーケンシャル / 非差分
+        0xCA, # 算術 / プログレッシブ / 非差分
+        0xCB, # 算術 / ロスレス / 非差分
+        0xCD, # 算術 / シーケンシャル / 差分
+        0xCE, # 算術 / プログレッシブ / 差分
+        0xCF, # 算術 / ロスレス / 差分
+    ]
+
+    RST = [
+        0xD0,
+        0xD1,
+        0xD2,
+        0xD3,
+        0xD4,
+        0xD5,
+        0xD6,
+        0xD7,
+    ]
+
+    # cf. https://hp.vector.co.jp/authors/VA032610/JPEGFormat/JPEGsegment.htm
+    DQT = 0xDB # 量子化テーブル定義
+    DHT = 0xC4 # ハフマンテーブル定義
+    DRI = 0xDD
+    COM = 0xFE
+    SOS = 0xDA
+    APP = [
+        0xE0,
+        0xE1,
+        0xE2,
+        0xE3,
+        0xE4,
+        0xE5,
+        0xE6,
+        0xE7,
+        0xE8,
+        0xE9,
+        0xEA,
+        0xEB,
+        0xEC,
+        0xED,
+        0xEE,
+        0xEF,
+    ]
+
+    REGISTERED_MARKERS = []
+    REGISTERED_MARKERS.extend(SOF)
+    REGISTERED_MARKERS.extend(APP)
+    REGISTERED_MARKERS.extend(RST)
+    REGISTERED_MARKERS.append(SOI)
+    REGISTERED_MARKERS.append(DQT)
+    REGISTERED_MARKERS.append(DHT)
+    REGISTERED_MARKERS.append(DRI)
+    REGISTERED_MARKERS.append(COM)
+    REGISTERED_MARKERS.append(SOS)
+    REGISTERED_MARKERS.append(EOI)
+    REGISTERED_MARKERS.append(JUST_FF)

segments/APP.py

@@ -0,0 +1,50 @@
+import sys
+import os
+import warnings
+sys.path.append(os.pardir)
+
+from segmentsCommon import SegmentsCommon
+from markers import Markers
+from APPs.APP0 import APP0
+
+class APP(SegmentsCommon):
+    def __init__(self):
+        super(APP, self).__init__()
+
+    def analyze(self, marker, body):
+        if marker == Markers.APP[0]:
+            return APP0().analyze(marker, body)
+        elif marker == Markers.APP[1]:
+            pass
+        elif marker == Markers.APP[2]:
+            pass
+        elif marker == Markers.APP[3]:
+            pass
+        elif marker == Markers.APP[4]:
+            pass
+        elif marker == Markers.APP[5]:
+            pass
+        elif marker == Markers.APP[6]:
+            pass
+        elif marker == Markers.APP[7]:
+            pass
+        elif marker == Markers.APP[8]:
+            pass
+        elif marker == Markers.APP[9]:
+            pass
+        elif marker == Markers.APP[10]:
+            pass
+        elif marker == Markers.APP[11]:
+            pass
+        elif marker == Markers.APP[12]:
+            pass
+        elif marker == Markers.APP[13]:
+            pass
+        elif marker == Markers.APP[14]:
+            pass
+        elif marker == Markers.APP[15]:
+            pass
+        else:
+            raise ValueError("invailed marker. marker " + marker.to_bytes(1, 'big') + " found. class APP is not excepting it.\nAPP expecting marker is below.\n" + str([m.to_bytes(1, 'big') for m in Markers.APP]))
+        # raise NotImplementedError("this type APP segment not supported. please inform to the author.")
+        warnings.warn("this type APP segment not supported. please inform to the author.")

segments/APPs/APP0.py

@@ -0,0 +1,27 @@
+import sys
+import os
+sys.path.append(os.pardir)
+sys.path.append("../..")
+
+from segmentsCommon import SegmentsCommon
+from markers import Markers
+
+class APP0(SegmentsCommon):
+    def __init__(self):
+        super(APP0, self).__init__()
+
+    def analyze(self, marker, body):
+        analyzed = {}
+        analyzed["segmentName"] = "APP0"
+        assert body[0:5] == [ord('J'), ord('F'), ord('I'), ord('F'), 0x00], 'for APP0, JFIF is only supported. maybe it contains JFXX segment?'
+        # TODO: support JFXX segent
+        analyzed["format"] = "JFIF"
+        analyzed["ver"] = str(body[5]) + '.' + str(body[6])
+        analyzed["units"] = body[7] # 0 = None, 1 = dpi, 2 = dpcm
+        analyzed["xdensity"] = body[8] * 0x100 + body[9]
+        analyzed["ydensity"] = body[10] * 0x100 + body[11]
+        analyzed["xthumbnail"] = body[12]
+        analyzed["ythumbnail"] = body[13]
+        if analyzed["xthumbnail"] * analyzed["ythumbnail"] > 0:
+            analyzed["RGB"] = body[14:14 + 3 * analyzed["xthumbnail"] * analyzed["ythumbnail"]]
+        return analyzed

segments/COM.py

@@ -0,0 +1,18 @@
+import sys
+import os
+sys.path.append(os.pardir)
+
+from segmentsCommon import SegmentsCommon
+from markers import Markers
+
+class COM(SegmentsCommon):
+    def __init__(self):
+        super(COM, self).__init__()
+
+    def analyze(self, marker, body):
+        if marker != Markers.COM:
+            raise ValueError("invailed marker. class COM excepting marker is " + Markers.COM.to_bytes(1, 'big') + ". but " + marker.to_bytes(1, 'big') + " found.")
+        analyzed = {}
+        analyzed["segmentName"] = "COM"
+        analyzed["comment"] = ''.join([chr(i) for i in body])
+        return analyzed