From 8dbbe294b9a8071b6d0672db826bc7dec585b939 Mon Sep 17 00:00:00 2001
From: Derrick Chambers <chambers.ja.derrick@gmail.com>
Date: Fri, 6 Dec 2024 17:34:24 -0800
Subject: [PATCH] make segyio optional

---
 dascore/io/segy/__init__.py |  7 +++++
 dascore/io/segy/core.py     | 14 +++++----
 dascore/io/segy/utils.py    | 58 ++++++++++++++++++++++++++++++++++++-
 3 files changed, 72 insertions(+), 7 deletions(-)

diff --git a/dascore/io/segy/__init__.py b/dascore/io/segy/__init__.py
index 2a421dd7..e6f8a3f9 100644
--- a/dascore/io/segy/__init__.py
+++ b/dascore/io/segy/__init__.py
@@ -5,6 +5,13 @@
 -----
 - Distance information is not found in most SEGY DAS files so returned
   dimensions are "channel" and "time" rather than "distance" and "time".
+- Segy standards found at: https://library.seg.org/pb-assets/technical-standards
+
+segy v1 spec: seg_y_rev1-1686080991247.pdf
+
+segy v2 spec: seg_y_rev2_0-mar2017-1686080998003.pdf
+
+segy v2.1 spec: seg_y_rev2_1-oct2023-1701361639333.pdf
 
 Examples
 --------
diff --git a/dascore/io/segy/core.py b/dascore/io/segy/core.py
index 5e82c5bf..b892e53b 100644
--- a/dascore/io/segy/core.py
+++ b/dascore/io/segy/core.py
@@ -7,7 +7,7 @@
 import dascore as dc
 from dascore.io.core import FiberIO
 
-from .utils import _get_attrs, _get_coords, _get_filtered_data_and_coords
+from .utils import _get_attrs, _get_coords, _get_filtered_data_and_coords, _is_segy
 
 
 class SegyV2(FiberIO):
@@ -21,11 +21,13 @@ class SegyV2(FiberIO):
 
     def get_format(self, path, **kwargs) -> tuple[str, str] | bool:
         """Make sure input is segy."""
-        try:
-            with segyio.open(path, ignore_geometry=True):
-                return self.name, self.version
-        except Exception:
-            return False
+        with open(path, "rb") as fp:
+            return _is_segy(fp)
+        # try:
+        #     with segyio.open(path, ignore_geometry=True):
+        #         return self.name, self.version
+        # except Exception:
+        #     return False
 
     def read(self, path, time=None, channel=None, **kwargs):
         """
diff --git a/dascore/io/segy/utils.py b/dascore/io/segy/utils.py
index 97f4e75e..89eae603 100644
--- a/dascore/io/segy/utils.py
+++ b/dascore/io/segy/utils.py
@@ -4,13 +4,69 @@
 
 import datetime
 
+# --- Getting format/version
 import numpy as np
 from segyio import TraceField
 
 import dascore as dc
 from dascore.core import get_coord_manager
 
-# --- Getting format/version
+# Valid data format codes as specified in the SEGY rev1 manual.
+VALID_FORMATS = [1, 2, 3, 4, 5, 8]
+
+# This is the maximum possible interval between two samples due to the nature
+# of the SEG Y format.
+MAX_INTERVAL_IN_SECONDS = 0.065535
+
+# largest number possible with int16
+MAX_NUMBER_OF_SAMPLES = 32767
+
+
+def twos_comp(bytes_):
+    """Get twos complement of bytestring."""
+    bits = len(bytes_) * 8
+    val = int.from_bytes(bytes_, "big")
+    if (val & (1 << (bits - 1))) != 0:  # if sign bit is set e.g., 8bit: 128-255
+        val = val - (1 << bits)  # compute negative value
+    return val  # return positive value as is
+
+
+def _is_segy(fp):
+    """
+    Return True if file pointer contains segy formatted data.
+
+    Based on ObsPy's implementation writen by Lion Krischer.
+    https://github.com/obspy/obspy/blob/master/obspy/io/segy/core.py
+    """
+    # # Read 400byte header into byte string.
+    # fp.seek(3200)
+    # header = fp.read(400)
+    # data_trace_count = twos_comp(header[12:14])
+    # auxiliary_trace_count = twos_comp(header[14:16])
+    # sample_interval = twos_comp(header[16:18])
+    # samples_per_trace = twos_comp(header[20:22])
+    # data_format_code = twos_comp(header[24:26])
+    # format_number_major = int.from_bytes(header[300:301])
+    # format_number_minor = int.from_bytes(header[301:302])
+    # fixed_len_flag = twos_comp(header[302:304])
+    #
+    #
+    # if _format_number not in (0x0000, 0x0100, 0x0010, 0x0001):
+    #     return False
+    #
+    # _fixed_length = unpack(fmt, _fixed_length)[0]
+    # _extended_number = unpack(fmt, _extended_number)[0]
+    # # Make some sanity checks and return False if they fail.
+    # if (
+    #     _sample_interval <= 0
+    #     or _samples_per_trace <= 0
+    #     or _number_of_data_traces < 0
+    #     or _number_of_auxiliary_traces < 0
+    #     or _fixed_length < 0
+    #     or _extended_number < 0
+    # ):
+    #     return False
+    return True
 
 
 def _get_filtered_data_and_coords(segy_fi, coords, time=None, channel=None):