Files
A-Team-Security-Infra-Agent…/backend/services/terraform_exec.py
nogueiraguh 1135e9d6a9 refactor: decompose monolith — app.py 10,600→143 lines, 30+ modules
Fase 0 complete: extract all business logic, auth, database, and 176
endpoints from monolithic app.py into dedicated modules.

Structure:
- app.py: FastAPI hub (CORS, routers, startup/shutdown)
- models.py: 13 Pydantic request models
- utils.py: shared utilities (embed status, upload validation, process registries)
- config.py: constants, env vars, model catalogs
- auth/: crypto, jwt_auth, oidc, rate_limit
- database/: db(), init_db(), schema DDL
- services/: genai, compliance, chat, terraform, embeddings, cis_reports, mcp
- routes/: 13 APIRouter modules (auth, users, oci_config, oci_explorer,
  genai, mcp, adb, embeddings, reports, chat, terraform, settings, cis_engine)

Also: README updated with OCIR auth instructions for manual docker run.

86 tests passing.
2026-04-06 15:20:10 -03:00

422 lines
19 KiB
Python

"""Terraform execution — split monolith, write files, plan/apply/destroy."""
import os, json, uuid, asyncio, subprocess, re
from pathlib import Path
from datetime import datetime
from config import TERRAFORM_DIR, OCI_DIR, log
from database import db
from auth.crypto import _safe_dec
from auth.jwt_auth import _audit, _verify_config_access
def _split_tf_monolith(content: str) -> dict:
"""Split a monolithic HCL string into multiple logical files by resource type.
Returns dict of {filename: content} or None if splitting not needed."""
import re as _re
content = _fix_single_line_blocks(content)
lines = content.split('\n')
top_blocks = []
preamble = []
accum = []
in_block = False
b_depth = 0
cur_header = ''
for line in lines:
stripped = _re.sub(r'"[^"]*"', '', line)
opens = stripped.count('{')
closes = stripped.count('}')
if not in_block:
hdr = _re.match(r'^\s*(variable|output|resource|data|provider|module)\s+"', line) or \
_re.match(r'^\s*(locals|terraform)\s*\{', line)
if hdr:
in_block = True
cur_header = line
accum = preamble + [line]
preamble = []
b_depth = opens - closes
if b_depth <= 0:
b_depth = 0
if b_depth == 0 and opens > 0:
top_blocks.append((cur_header, '\n'.join(accum)))
in_block = False
accum = []
else:
preamble.append(line)
else:
accum.append(line)
b_depth += opens - closes
if b_depth <= 0:
b_depth = 0
top_blocks.append((cur_header, '\n'.join(accum)))
in_block = False
accum = []
cur_header = ''
if accum:
top_blocks.append((cur_header or '', '\n'.join(accum)))
if len(top_blocks) < 3:
return None
cats = {'variables': [], 'outputs': [], 'networking': [], 'compute': [], 'database': [],
'firewall': [], 'loadbalancer': [], 'storage': [], 'iam': [], 'drg': [],
'data': [], 'providers': [], 'other': []}
for header, body in top_blocks:
h = header.strip()
if h.startswith('variable '):
cats['variables'].append(body)
elif h.startswith('output '):
cats['outputs'].append(body)
elif h.startswith('provider '):
cats['providers'].append(body)
elif _re.search(r'resource\s+"oci_core_(drg|remote_peering|drg_route|drg_attachment)', h):
cats['drg'].append(body)
elif _re.search(r'resource\s+"oci_core_(vcn|subnet|internet_gateway|nat_gateway|service_gateway|route_table|security_list|network_security_group|dhcp_options|local_peering_gateway|virtual_circuit)', h):
cats['networking'].append(body)
elif _re.search(r'resource\s+"oci_core_instance', h):
cats['compute'].append(body)
elif _re.search(r'resource\s+"oci_(database|nosql|mysql|psql)', h) or _re.search(r'resource\s+"oci_core_autonomous', h):
cats['database'].append(body)
elif _re.search(r'resource\s+"oci_network_firewall', h):
cats['firewall'].append(body)
elif _re.search(r'resource\s+"oci_(load_balancer|network_load_balancer)', h):
cats['loadbalancer'].append(body)
elif _re.search(r'resource\s+"oci_(objectstorage|file_storage)', h):
cats['storage'].append(body)
elif _re.search(r'resource\s+"oci_identity', h):
cats['iam'].append(body)
elif h.startswith('data '):
cats['data'].append(body)
else:
cats['other'].append(body)
result = {}
mapping = [('variables.tf', 'variables'), ('providers.tf', 'providers'), ('networking.tf', 'networking'),
('drg.tf', 'drg'), ('compute.tf', 'compute'), ('database.tf', 'database'),
('firewall.tf', 'firewall'), ('loadbalancer.tf', 'loadbalancer'), ('storage.tf', 'storage'),
('iam.tf', 'iam'), ('data.tf', 'data'), ('main.tf', 'other'), ('outputs.tf', 'outputs')]
for fname, key in mapping:
if cats[key]:
result[fname] = '\n\n'.join(cats[key])
return result if len(result) >= 3 else None
def _write_tf_files(wdir: Path, tf_code: str):
"""Parse tf_code for '// filename: xxx.tf' markers and write separate files.
If only 1-2 large files, auto-split into multiple files by resource type."""
import re as _re
parts = _re.split(r'^//\s*filename:\s*(\S+)\s*$', tf_code, flags=_re.MULTILINE)
files = {}
if len(parts) >= 3:
if parts[0].strip():
files['main.tf'] = parts[0].strip()
for i in range(1, len(parts), 2):
fname = parts[i].strip().replace("/", "_").replace("..", "_")
content = parts[i + 1].strip() if i + 1 < len(parts) else ""
if fname and content:
files[fname] = content
else:
files['main.tf'] = tf_code
# Auto-split: if 1-2 large files, split by resource type
if len(files) <= 2:
all_content = '\n\n'.join(files.values())
if len(all_content) > 2000:
split = _split_tf_monolith(all_content)
if split:
files = split
log.info(f"Backend auto-split monolithic TF into {len(files)} files: {list(files.keys())}")
# Deduplicate: if a resource appears in multiple files, keep only in the split file
# This handles edge cases where monolithic + split files both exist
if len(files) > 2:
resource_locations = {} # "type.name" -> [(filename, line)]
dupes_in = set()
for fname, content in files.items():
for m in _re.finditer(r'^resource\s+"(\S+)"\s+"(\S+)"', content, _re.MULTILINE):
key = f'{m.group(1)}.{m.group(2)}'
resource_locations.setdefault(key, []).append(fname)
# Find files that contain ONLY duplicate resources (= monolithic leftovers)
for key, fnames in resource_locations.items():
if len(fnames) > 1:
# The largest file is likely the monolithic one
largest = max(fnames, key=lambda f: len(files.get(f, '')))
dupes_in.add(largest)
# Remove monolithic files that are fully duplicated
for fname in dupes_in:
if all(
any(f2 != fname for f2 in resource_locations.get(key, []))
for key, flist in resource_locations.items()
if fname in flist
) and len(files) - 1 >= 2:
log.info(f"Removing duplicate monolithic file: {fname}")
del files[fname]
# Fix single-line blocks in all files
for fname in files:
files[fname] = _fix_single_line_blocks(files[fname])
# Write files to disk
for fname, content in files.items():
(wdir / fname).write_text(content)
async def _terraform_exec(wid: str, action: str, user: dict):
"""Background: run terraform init + plan/apply/destroy in workspace dir."""
log.info(f"Terraform {action} started: wid={wid}")
status_col = f"{action}_output"
final_ok = {"plan": "planned", "apply": "applied", "destroy": "destroyed"}[action]
try:
with db() as c:
ws = c.execute("SELECT * FROM terraform_workspaces WHERE id=?", (wid,)).fetchone()
if not ws:
return
wdir = TERRAFORM_DIR / wid
wdir.mkdir(parents=True, exist_ok=True)
# Clean old .tf files (keep .terraform/, state, lock)
for old_tf in wdir.glob("*.tf"):
old_tf.unlink()
# Write .tf files — parse // filename: comments to split into separate files
_write_tf_files(wdir, ws["tf_code"] or "")
# Auto-generate provider.tf from OCI config
# Detect provider aliases declared by the model in generated files
import re as _re2
oci_cfg = _get_oci_config(ws["oci_config_id"])
alias_blocks = [] # list of (alias_name, region_ref)
# Scan all .tf files for provider "oci" { alias = "..." ... region = ... }
provider_block_re = _re2.compile(r'provider\s+"oci"\s*\{', _re2.MULTILINE)
for tf_file in sorted(wdir.glob("*.tf")):
if tf_file.name == "provider.tf":
continue
content = tf_file.read_text()
# Find each provider "oci" block and extract alias + region
blocks_to_remove = []
for m in provider_block_re.finditer(content):
start = m.start()
# Find matching closing brace
depth = 0
end = start
for ci in range(m.end() - 1, len(content)):
if content[ci] == '{':
depth += 1
elif content[ci] == '}':
depth -= 1
if depth == 0:
end = ci + 1
break
block = content[start:end]
alias_m = _re2.search(r'alias\s*=\s*"([^"]+)"', block)
region_m = _re2.search(r'region\s*=\s*(.+)', block)
if alias_m:
alias_name = alias_m.group(1)
region_ref = region_m.group(1).strip().rstrip("}").strip() if region_m else '"unknown"'
alias_blocks.append((alias_name, region_ref))
blocks_to_remove.append((start, end))
# Remove model-generated provider blocks (we centralize in provider.tf)
if blocks_to_remove:
new_content = content
for s, e in reversed(blocks_to_remove):
new_content = new_content[:s] + new_content[e:]
# Also remove leading comments right before removed blocks
new_content = _re2.sub(r'\n(//[^\n]*\n){1,5}\s*\n', '\n\n', new_content)
new_content = new_content.strip()
if new_content:
tf_file.write_text(new_content + "\n")
else:
tf_file.unlink() # Remove empty file
# Scan all .tf files for variable declarations with region-like defaults
# so we can map alias providers to the correct variable reference
var_region_map = {} # variable_name -> default_value
var_re = _re2.compile(r'variable\s+"([^"]*region[^"]*)"\s*\{', _re2.IGNORECASE)
for tf_file in sorted(wdir.glob("*.tf")):
if tf_file.name in ("provider.tf", "terraform.tfvars"):
continue
content = tf_file.read_text()
for vm in var_re.finditer(content):
var_region_map[vm.group(1)] = f'var.{vm.group(1)}'
# Check if variable "region" is declared — use it for primary provider
has_region_var = "region" in var_region_map
# Also scan for provider refs in resource blocks (provider = oci.xxx)
for tf_file in sorted(wdir.glob("*.tf")):
if tf_file.name == "provider.tf":
continue
content = tf_file.read_text()
for ref_m in _re2.finditer(r'provider\s*=\s*oci\.(\w+)', content):
ref_alias = ref_m.group(1)
if ref_alias not in [a[0] for a in alias_blocks]:
# Try to find a matching region variable for this alias
# e.g. alias "mad_3" might match var.region_secondary
region_ref = 'var.region' # fallback to primary region var
for vname, vref in var_region_map.items():
if vname != "region": # prefer non-primary region vars for aliases
region_ref = vref
break
alias_blocks.append((ref_alias, region_ref))
passphrase = oci_cfg.get('pass_phrase', '')
cred_block = f''' tenancy_ocid = "{oci_cfg.get('tenancy', '')}"
user_ocid = "{oci_cfg.get('user', '')}"
fingerprint = "{oci_cfg.get('fingerprint', '')}"
private_key_path = "{oci_cfg.get('key_file', '')}"'''
if passphrase:
cred_block += f'\n private_key_password = "{passphrase}"'
# Primary region: MUST use var.region — model is required to declare it
with db() as c:
oci_row = c.execute("SELECT region FROM oci_configs WHERE id=?", (ws["oci_config_id"],)).fetchone()
primary_region = oci_row["region"] if oci_row else oci_cfg.get("region", "")
if not has_region_var:
error_msg = (
f'ERRO: O código Terraform não declarou variable "region". '
f'Isso é obrigatório para garantir que os recursos sejam provisionados na região correta. '
f'Adicione no variables.tf: variable "region" {{ type = string; default = "{primary_region}" }}'
)
log.warning(f"Workspace {wid}: missing variable 'region' — blocking plan")
with db() as c:
c.execute("UPDATE terraform_workspaces SET status='failed', plan_output=?, error=?, updated_at=datetime('now') WHERE id=?",
(error_msg, error_msg, wid))
return
region_value = 'var.region'
provider_tf = f'''terraform {{
required_providers {{
oci = {{
source = "oracle/oci"
}}
}}
}}
provider "oci" {{
{cred_block}
region = {region_value}
}}
'''
# Add alias providers with same credentials
seen_aliases = set()
for alias_name, region_ref in alias_blocks:
if alias_name in seen_aliases:
continue
seen_aliases.add(alias_name)
provider_tf += f'''
provider "oci" {{
alias = "{alias_name}"
{cred_block}
region = {region_ref}
}}
'''
(wdir / "provider.tf").write_text(provider_tf)
# Auto-generate terraform.tfvars — scan declared variables and provide values from OCI config
with db() as c:
oci_row = c.execute("SELECT compartment_id, region, ssh_public_key FROM oci_configs WHERE id=?", (ws["oci_config_id"],)).fetchone()
comp_id = ws["compartment_id"] if ws["compartment_id"] else (_safe_dec(oci_row["compartment_id"]) if oci_row and oci_row["compartment_id"] else "")
try:
ssh_pub_key = oci_row["ssh_public_key"] if oci_row else ""
except (IndexError, KeyError):
ssh_pub_key = ""
ssh_pub_key = ssh_pub_key or ""
# Scan all declared variables in .tf files
declared_vars = set()
for tf_file in sorted(wdir.glob("*.tf")):
if tf_file.name in ("provider.tf", "terraform.tfvars"):
continue
content = tf_file.read_text()
for vm in _re2.finditer(r'variable\s+"([^"]+)"', content):
declared_vars.add(vm.group(1))
# Map known variable names to OCI config values
var_values = {"compartment_id": comp_id}
oci_var_map = {
"tenancy_ocid": oci_cfg.get("tenancy", ""),
"tenancy_id": oci_cfg.get("tenancy", ""),
"user_ocid": oci_cfg.get("user", ""),
"fingerprint": oci_cfg.get("fingerprint", ""),
"private_key_path": oci_cfg.get("key_file", ""),
"ssh_public_key": ssh_pub_key,
"ssh_authorized_keys": ssh_pub_key,
}
# NOTE: "region" is intentionally NOT included — Terraform uses the default from variable declarations
for vname, vval in oci_var_map.items():
if vname in declared_vars and vval:
var_values[vname] = vval
tfvars_lines = [f'{k} = "{v}"' for k, v in var_values.items()]
(wdir / "terraform.tfvars").write_text("\n".join(tfvars_lines) + "\n")
output_lines = []
def _update_output(text):
output_lines.append(text)
with db() as c:
c.execute(f"UPDATE terraform_workspaces SET {status_col}=?, updated_at=datetime('now') WHERE id=?",
("\n".join(output_lines[-200:]), wid))
# terraform init
_update_output("$ terraform init ...")
proc_init = await asyncio.create_subprocess_exec(
"terraform", f"-chdir={wdir}", "init", "-no-color", "-input=false",
stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.STDOUT)
while True:
line = await proc_init.stdout.readline()
if not line:
break
_update_output(line.decode(errors="replace").rstrip())
await proc_init.wait()
if proc_init.returncode != 0:
_update_output(f"\n❌ terraform init failed (exit {proc_init.returncode})")
with db() as c:
c.execute("UPDATE terraform_workspaces SET status='failed', error=?, updated_at=datetime('now') WHERE id=?",
("terraform init failed", wid))
return
# terraform action
_update_output(f"\n$ terraform {action} ...")
cmd = ["terraform", f"-chdir={wdir}", action, "-no-color", "-input=false"]
if action in ("apply", "destroy"):
cmd.append("-auto-approve")
proc = await asyncio.create_subprocess_exec(
*cmd, stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.STDOUT)
_running_terraform[wid] = proc
while True:
line = await proc.stdout.readline()
if not line:
break
_update_output(line.decode(errors="replace").rstrip())
await proc.wait()
_running_terraform.pop(wid, None)
if proc.returncode == 0:
output_lines.append(f"\n✅ terraform {action} completed successfully")
full_output = "\n".join(output_lines)
with db() as c:
c.execute(f"UPDATE terraform_workspaces SET {status_col}=?, status=?, updated_at=datetime('now') WHERE id=?",
(full_output, final_ok, wid))
_audit(user["id"], user["username"], f"terraform_{action}", wid, f"status={final_ok}")
else:
output_lines.append(f"\n❌ terraform {action} failed (exit {proc.returncode})")
full_output = "\n".join(output_lines)
with db() as c:
c.execute(f"UPDATE terraform_workspaces SET {status_col}=?, status='failed', error=?, updated_at=datetime('now') WHERE id=?",
(full_output, f"terraform {action} failed (exit {proc.returncode})", wid))
except Exception as e:
log.error(f"Terraform exec error: {e}")
with db() as c:
c.execute("UPDATE terraform_workspaces SET status='failed', error=?, updated_at=datetime('now') WHERE id=?",
(str(e)[:500], wid))
# ── Audit ─────────────────────────────────────────────────────────────────────